• occupational exposure
• lung cancer
• health screening

Lung cancer is responsible for nearly one in five of all cancer deaths in the world and is the leading occupationally related cancer type.1 Tobacco smoke exposure contributes the most significant risk although in developed countries, occupational exposures are estimated to contribute to 10%–30% of all lung cancers.1 The International Agency for Research on Cancer recognises at least 13 occupational exposures that are associated with a raised risk of lung cancer and at least 6 of these (asbestos, arsenic, radon, polyaromatic hydrocarbons, silica and nickel) may have a more than additive (ie, synergistic) risk when combined with tobacco smoke exposure.1

The National Lung Screen Trial (NLST) demonstrated a 20.0% mortality reduction for lung cancer deaths with three screening rounds using low-dose CT (LDCT) of the chest.2 However, LDCT screening for lung cancer is only effective when a high-risk population is screened. The number of lung cancer deaths prevented by LDCT is strongly related to the underlying risk of lung cancer in the population. The NLST used criteria of aged 55–74 years, at least 30 pack-year history and <15 years since quitting to identify the population at risk. The result from NLST is more remarkable considering that the lowest quintile of risk had just 1 lung cancer death prevented, compared with 33 in the highest risk quintile.3 This suggests that the criteria used by the NLST (now adapted by the United States Preventative Services Task Force) could be improved to enhance the pretest probability of lung cancer in the population being screened. In addition to age and tobacco smoke exposure, probabilistic lung cancer risk models account for factors such as chronic obstructive pulmonary disease (COPD) and family history of lung cancer.4 The presence of asbestosis is known to increase the risk of lung cancer and pleural plaques are likely to be a marker of asbestos exposure only.5 To date, few lung cancer risk models account for occupational exposures and there is certainly no consensus as to the optimal approach in doing so.

The National Comprehensive Cancer Network (NCCN) guidelines have recognised the contribution of occupational exposure(s) and reduced the required tobacco smoke exposure to 20 pack-years to be eligible for screening in the context of such exposures.6 This reduction is arbitrary but of great import—it highlights the significance of occupational exposures in the risk of lung cancer and that there are currently few data to support the use of semiquantitative prediction models to account for occupational exposures and lung cancer risk.

The paper by Welch in this issue of OEM provides further evidence for the importance of occupational exposures in risk assessment for lung cancer screening.7 The authors describe the results of an early lung cancer detection programme using LDCT on a population of ex-construction workers who fitted the modified NCCN criterion with at least 5 years of potentially multiple occupational lung carcinogen exposures. Subjects were also included with < 5 years of exposure if there was evidence of COPD, asbestosis or pleural plaques. Using these simple criteria, the authors recruited a population of 1290 participants with a baseline prevalence of 1.6% for lung cancer within the cohort; this is comparable to the NLST population2 and is complementary to other reports from asbestos-exposed populations.8 9 While these reports provide supportive evidence for the need to further and adequately account for occupational exposures in lung cancer risk assessment, the reports also highlight the challenges in doing so. For instance, Welch used 5 years of exposure(s) to qualify for screening and, by contrast, the Western Australian-based Asbestos Review programme uses 3 months cumulative full-time exposure to asbestos to join the programme.9 Attempts to quantify occupational carcinogen exposures using job exposure matrices10 are important, although their complexity and potential weaknesses highlight the inherent challenges in epidemiological terms of estimating exposures to workers that occurred many decades ago, across many hundreds of different jobs.

The relationship of asbestos exposure and risk of lung cancer continues to have variable relationships described. For example, in workers exposed to crocidolite, a clear exposure–response relationship of cumulative and duration of asbestos exposure was demonstrated.11 The risk of lung cancer with combined tobacco and asbestos exposure is at least additive, but less than multiplicative. More recently, data in workers exposed to mixed asbestos fibres suggested a less clear relationship.12 Any estimate of occupational exposure needs to be simple and easily assessed to be adopted and widely utilised as part of a screening programme; it is clear that our current understanding even with asbestos (probably the best studied occupational lung carcinogen) does not provide the clarity required. Perhaps the duration of exposure provides a more pragmatic and practical estimate of carcinogen exposure, and the paper by Welch provides further support for this notion.

Aside from how to identify a suitable occupationally exposed cohort, the downstream cost of incidental findings, overdiagnosis, how to optimally manage indeterminate nodules and the vital need to ensure further radiation exposure is as low as reasonably practicable (in a population already exposed to at least two carcinogens) represent further requirements and challenges before more widespread adoption of LDCT screening programmes in occupational cohorts. We should also not lose sight of the paramount importance of primary lung cancer prevention by workplace exposure-reduction and smoking cessation measures.

Given the significant contribution of occupational exposures to the risk of lung cancer and that LDCT lung cancer screening is only effective in high-risk populations, there is a clear need to define suitably high-risk populations to be screened. At this time, there are no generalisable models to provide a semiquantitative exposure estimate for the multiple relevant occupational carcinogens to inform selection for lung cancer screening programmes. Efforts should continue towards this. In the meantime, Welch has provided an important contribution to the growing case to more routinely consider and include occupational exposures (regardless of exactly which) as part of the selection for LDCT screening. For now, within carefully defined cohorts, perhaps a pragmatic estimate of occupational exposures using time as the exposure variable is adequate to identify suitably high-risk populations for lung cancer screening.