Abstract
This article summarises evidence of radiation effects and dose response associations derived from occupational studies based on a presentation delivered at the ICRP 2023 symposium. First, evidence derived from studies of uranium miners is reviewed, and then findings from studies of workers in the nuclear industry are discussed. Uranium miner studies provide strong evidence of excess lung cancer; the exposure–response association for that association appears quite linear and is somewhat steeper when using more contemporary data. Nuclear worker studies provide evidence of a dose–response association with the broader group of solid cancers; pooling of occupational cohorts allows investigators to extend and support these investigations.
URANIUM MINERS
The increase in demand for uranium following World War II brought attention to epidemiological research on miners. Exposures in the early years of uranium mining resulted in large excesses of respiratory disease mortality. Early epidemiological publications described the health and mortality of uranium miners employed in the US Colorado Plateau and Czech mines (Tomasek et al., 1994; Hornung and Meinhardt, 1987). Two highly influential reports, known as BEIR IV and BEIR VI, summarised evidence and provided pooled analyses of lung cancer among miners that included cohorts of uranium, as well as tin, iron, and fluorspar (National Research Council (U.S.) et al., 1988, National Research Council and Committee on Health Risks of Exposure to Radon (BEIR VI), 1999). Epidemiological cohort studies of uranium miners have come to play a major role in establishing that exposure to radon progeny causes lung cancer.
A recent extension and expansion of those efforts is the PUMA study, which is a collaborative effort of Canadian, German, French, Czech, and US investigators (Rage et al., 2020). PUMA was motivated by the increase in information since the BEIR VI analysis coming from updates of many of the national cohorts and the notable new German cohort: WISMUT. PUMA encompasses over 4 million person-years and nearly 52,000 deaths among over 100,000 uranium miners (Kelly-Reif et al., 2023). It was also motivated by scientific questions that would be difficult to address in individual cohorts, including questions about what effects other than lung cancer might be observed (Rage et al., 2020). All-cause mortality was 5% in excess, and cancer mortality was 23% in excess, due primarily to excess lung cancer (which was almost twice what was expected based on general population rates) (Richardson et al., 2020). Other cancer excesses include cancer of the stomach, liver, and larynx. Table 1 compares the results of PUMA to the BEIR VI model for the effect of radon progeny on lung cancer, which suggests that the radon–lung cancer association may vary with time since exposure, attained age, and exposure rate. The table reports the summary excess relative rate of lung cancer (ERR) per 100 working level months (WLM), showing that the PUMA estimate is similar to, but somewhat smaller in magnitude than, the estimate from the BEIR VI model. The next rows show that temporal factors are similar to those reported from the BEIR VI model: the ERR per 100 WLM diminishes with time since exposure, attained age, and exposure rate. The bottom row documents the increase in information (in terms of number of deaths and person-years) in the PUMA study as compared to the BEIR VI report.
PUMA model and BEIR VI exposure–age–concentration model for the association between cumulative radon exposure and lung cancer mortality with effect modification by time since exposure, attained age, and exposure rate. PUMA study of uranium miners and BEIR VI 11 miner study.
PUMA model and BEIR VI exposure–age–concentration model for the association between cumulative radon exposure and lung cancer mortality with effect modification by time since exposure, attained age, and exposure rate. PUMA study of uranium miners and BEIR VI 11 miner study.
Early studies of underground miners depended heavily upon information on deaths among miners employed in the early years of operations when exposures were high and tended to be poorly estimated. We undertook an analysis of miners employed during the more contemporary period of the uranium mining industry when radon exposures tended to be comparatively low (more comparable to settings of primary concern in contemporary radiation protection) and individual exposure data are available (minimising longstanding concerns about measurement errors for people employed in the early years). Fig. 1 shows the relationship between cumulative radon exposure and excess relative rate of lung cancer among miners first employed in 1960 or later. The relative rate of lung cancer increased in a linear fashion with cumulative exposure. There is a bit of downward curvature, but a linear–quadratic model for the association led to negligible improvement in model goodness of fit compared to a simple linear model. Using these more contemporary data, we fit the BEIR VI model: The ERR per unit exposure is similar to, albeit slightly lower than, that of the BEIR VI model (Richardson et al., 2022).
Excess relative rate of lung cancer mortality (circles) and observed number of deaths (numbers), by categories of cumulative exposure to radon progeny. Linear model for the association between cumulative exposure and lung cancer (dashed line). PUMA study: males hired in 1960+.
Reports on cancer among workers in the nuclear industry began to appear later than the uranium miner studies. In the late 1970s and early 1980s, there began to appear publications on nuclear worker cohort studies, beginning with reports on the Hanford cohort (Mancuso et al., 1976; Brodsky, 1979; Gilbert and Petersen, 1985), followed by reports on the Oak Ridge cohort, UK cohorts, and Atomic Energy of Canada (Beral et al., 1985; Smith and Douglas, 1986; Wing et al., 1991; Gribbin et al., 1993). Initially this literature was primarily discussed by the radiation research community in a qualitative fashion; reviews of the literature were conducted, and conclusions were drawn about the coherence of the findings. Results from individual studies tended to be imprecise with confidence intervals that often spanned the null. By the late 1980s, pooled analyses were undertaken for some national data in US (Gilbert et al., 1989, 1993) and UK (Beral et al., 1985) studies. The International Agency for Research on Cancer (IARC) took a lead in international pooling of nuclear worker studies and development of a common core protocol (Cardis et al., 1995).
In recent years, one of the largest pooling efforts has involved collaborative studies among partners from France, the UK, and the USA. The International Nuclear Workers Study (INWORKS) was undertaken to strengthen direct assessments of the risks from low-dose, low–dose-rate exposure to penetrating forms of ionising radiation (Hamra et al., 2015). Criteria for selection of the study cohorts included completeness and quality of data, start of facility operations, and exposure primarily to high-energy low–linear energy transfer penetrating radiations. The study included 309,932 workers and encompassed 10.7 million person-years of follow-up, with 103,553 deaths, of which 31,009 deaths were due to cancer and 28,089 deaths were due to solid cancer. INWORKS quantified associations between cumulative radiation dose, based on personal dosimeters, and rates of cancer mortality (Richardson et al., 2015, 2023).
Table 2 summarises several dose–response associations in terms of the excess relative rate per Gy. On average, cancer mortality rates increased 53% per Gy radiation dose; after dropping leukaemias and lymphomas from the cancers, the magnitude of the estimated association is very slightly reduced. To indirectly assess potential confounding by smoking, we excluded lung cancers and examined whether the association persists; if the association was driven by smoking rather than radiation, then we would expect a large change upon dropping lung cancer, but we don’t see that. The last row shows the association between cumulative radiation dose and chronic obstructive pulmonary disease, an outcome not known to be associated with low-dose ionising radiation but strongly associated with tobacco smoking. We observed minimal evidence of association between cumulative radiation dose and chronic obstructive pulmonary disease providing further indirect evidence that the association was not driven by smoking.
Estimated excess relative rate of mortality per unit colon dose (ERR per Gy). INWORKS 1944–2016.
Estimated excess relative rate of mortality per unit colon dose (ERR per Gy). INWORKS 1944–2016.
Fig. 2 illustrates the fit of the model to the data. The straight line is the estimate of the average increase in solid cancer mortality rates with increasing dose. There is some evidence of downward curvature. The addition of a parameter for the square of cumulative dose led to a modest improvement in model goodness of fit compared to a simple linear model. If we consider the restricted range of 0–400-mGy cumulative dose or 0–200-mGy cumulative dose, the magnitude of the slope is somewhat steeper over these restricted dose ranges. Because our primary interest is in the effect of external exposure, we examined results in analyses restricted to the 84% of workers who were never flagged for incorporated radionuclides or internal monitoring (ERR = 0.82 per Gy; 90% CI: 0.46 to 1.22). Workers employed in the early years of this industry were hired into a new industry created during WWII. Large numbers of healthy males were selected out of the workforce by military conscription. Screening of the workforce raised questions about health-related selection among early hires and how this may have differed between technical and non-technical workers, with different baseline mortality patterns, confounded by early industrial hazards, as well as questions about early dosimeters and other radiation control practices. To address concerns about potential biases associated with the early years of operations, we examined the association between cumulative radiation dose and solid cancer mortality restricted to the workers hired in 1958 or later and restricted to the workers hired in 1965 or later. Upon excluding workers hired in the earliest years of operations, our estimate of the ERR per Gy for solid cancer was larger than the estimate derived from analysis of the full cohort.
Relative rate of solid cancer mortality by categories of cumulative dose and fitted line. INWORKS, 1944–2016.
Uranium miner studies have provided strong evidence of excess lung cancer. The exposure–response association for lung cancer among uranium miners appears quite linear and is somewhat steeper when using more contemporary data. Future work will continue to address questions regarding cancers at other sites for which current evidence is less certain. Nuclear worker studies provide evidence of association with the broader group of solid cancers. The dose–response association is quite linear, and again there is more to learn about site-specific cancer risks. Excess attributable cases in worker studies are quite small, given the typically low-dose distributions in these cohorts. Pooling of occupational cohorts, beyond providing conditions for statistical precision, allows investigators to focus on questions of bias: confounding, selection, and measurement error. Sensitivity analyses and formal statistical approaches that clarify identifying conditions for causal effects provide approaches to further address concerns about bias.