Lung cancer mortality associated with protracted low-dose occupational radiation exposures and smoking behaviors in U.S. radiologic technologists, 1983-2012

Abstract

In the Japanese atomic bomb survivors, risk of lung cancer has been shown to increase with greater acute exposure to ionizing radiation. Although similar findings have been observed in populations exposed to low-dose, protracted radiation, such studies lack information on cigarette smoking history, a potential confounder. In a cohort of 106,068 U.S. radiologic technologists, we examined the association between estimated cumulative lung absorbed dose from occupational radiation exposure and lung cancer mortality. Poisson regression models, adjusted for attained age, sex, birth cohort, pack-years smoked, and years since quitting smoking, were used to calculate linear excess relative risks (ERR) per 100 mGy, using time-dependent cumulative lung absorbed dose, lagged 10 years. Mean cumulative absorbed dose to the lung was 25 mGy (range: 0-810 mGy). During the 1983-2012 follow-up, 1,090 participants died from lung cancer. Greater occupational radiation lung dose was not associated with lung cancer mortality overall (ERR per 100 mGy: −0.02, 95% CI: <0-0.13). However, significant dose-response relationships were observed for some subgroups, which might be false positive results given the number of statistical tests performed. As observed in other studies of radiation and smoking, the interaction between radiation and smoking appeared to be sub-multiplicative with an ERR per 100 mGy of 0.41 (95% CI: 0.01-1.15) for those who smoked <20 pack-years and −0.03 (95% CI: <0-0.15) for those who smoked ≥20 pack-years. This study provides some evidence that greater protracted radiation exposure in the low-dose range is positively associated with lung cancer mortality.

INTRODUCTION

Lung cancer is a leading contributor to cancer incidence and mortality worldwide.¹ While most cases of lung cancer are attributed to cigarette smoking, 25% of cases occur in non-smokers², where occupational and environmental exposures account for the largest burden of disease.³ There is substantial experimental and epidemiologic evidence that ionizing radiation exposure increases lung cancer risk.⁴ A clear dose-response for lung cancer has been observed in Japanese atomic bomb survivors with a single, acute low-to-moderate exposure to radiation and in individuals with prolonged exposure to inhaled residential radon gas.^5,6 However, the risk of lung cancer following low-dose, protracted occupational exposures to external radiation is less well-understood due to the inability to properly control for the confounding effects of cigarette smoking in many of these studies.^7-10 A better understanding of these and other effects of low-dose, protracted radiation exposure would have important implications for the health of the approximately 23 million radiation workers worldwide.¹¹

In the United States Radiologic Technologists (USRT) Study, a greater self-reported number of years worked in the 1950s (when occupational exposures were much higher compared to more recent decades) and ever versus never working with nuclear medicine procedures (associated with higher annual badge dose readings) have been associated with greater risk of lung cancer death.^12,13 This is the first study using the USRT data to evaluate the occupational radiation dose-response relationship for lung cancer mortality using estimated cumulative lung absorbed doses. The USRT is also one of few cohorts of radiation workers in which it is possible to evaluate confounding and modifying effects of cigarette smoking on the association between cumulative radiation exposure and lung cancer mortality. In studies of Japanese atomic bomb survivors, there was evidence of an excess radiation-related risk of lung cancer for light-to-moderate smokers but not heavy smokers.^14,15 We hypothesize that the risk of lung cancer mortality increases with greater cumulative occupational radiation lung absorbed dose and that the magnitude of the dose-response differs by smoking status.

MATERIALS AND METHODS

Study population and data sources

The details of the study population and survey methodology have been previously published and are available online at http://radtechstudy.nci.nih.gov/.^16,17 In brief, the USRT cohort includes 146,021 current and former radiologic technologists who were certified by the American Registry of Radiologic Technologists (ARRT) for at least two years from 1926 to 1982. ARRT is a professional U.S. society that provides formal certification, which is not mandatory, but highly recommended. The study protocol is approved annually by the institutional review boards of the National Cancer Institute and the University of Minnesota.

Of 132,298 living and located radiologic technologists who received the first mailed questionnaire between 1983 and 1989, 68% responded. Between 1993 and 1998, 126,628 living and located radiologic technologists were mailed a second questionnaire (72% response rate). Cohort members who completed at least one of the first two questionnaires were mailed (or administered by telephone for technologists who first began working before 1950) a third (2003-2005) and fourth (2012-2014) questionnaire, with response rates of 72% and 62%, respectively. All four questionnaires included questions about work history practices, medical history (including diagnoses of specific types of cancer), smoking habits, and other lifestyle and socio-demographic factors.

The 110,368 radiologic technologists in the USRT who responded to either the first or second questionnaire (or both) were initially eligible for this study. Those with a previous history of cancer other than non-melanoma skin cancer before the date of entry into follow-up (N=4,242) or who had missing cancer history on the baseline questionnaire (N=58) were excluded. The final analytic sample comprised 106,068 radiologic technologists, including 80,180 women and 25,888 men.

Dose estimation

Complete details on the dosimetry system used for the USRT cohort have been published previously.^18,19 To summarize, estimates of absorbed dose to the lung for each participant from occupational radiation exposure were derived for each year worked from 1916-1997 by converting badge dose measurements or estimates to lung dose for each study subject. Before 1960, when individual badge dose data were not available, badge doses were estimated from historical period-specific distributions of badge dose measurements and work practice data derived from the literature. From 1960-1997, individual badge doses were used directly or estimated from a population badge dose distribution derived from 921,134 individual annual badge readings from 79,959 participants (72%). Information from three questionnaires on work history practices, including facility type, number of years worked, and types of radiologic procedures performed were used to estimate the technologists’ badge doses when individual badge measurements were not available. All badge dose measurements and estimates were converted to lung absorbed dose in a systematic fashion using dosimetric factors that reflected temporal changes in x-ray imaging technology, data from the third questionnaire on self-reported use of lead aprons across three decades (<1980, 1980-1989, ≥1990), literature data on lead apron thickness, and calculations of x-ray transmission through the aprons.¹⁹ One thousand annual cohort dose realizations were generated for the entire cohort using Monte Carlo simulation methods that accounted for shared and unshared errors and other uncertainties in the assumptions made about the available input data. The arithmetic means of the annual lung doses from the 1,000 realizations for each technologist in each year worked from 1916-1997 were used for these analyses.

Lung cancer ascertainment and follow-up

Technologists have been followed using the yearly recertification with the ARRT. Vital status of cohort members who were no longer recertified was ascertained through a linkage with the Social Security Administration and the National Death Index. Cause of death for decedents was obtained from the National Death Index. For the current analyses, the main outcome was defined as a recorded underlying cause of death due to cancer of the lung or bronchus following the Surveillance, Epidemiology, and End Results Program (SEER) cause-specific death coding (ICD8-162.1; ICD9-162.2-162.5, 162.8-162.9; ICD10-C34 (any)).²⁰ Participants were followed from the return date of the baseline questionnaire (e.g., the first completed of the first (1983-1989) or second questionnaire (1994-1998)), until the date of death, last known vital status, or December 31^st, 2012, whichever occurred first.

Smoking characteristics

Information on smoking behaviors was ascertained from responses to the first and second questionnaires. The smoking status of each participant was determined using time-dependent indicators. At each point in time over the follow-up period, participants were categorized as a never smoker, former smoker, current smoker, or unknown (if smoking information was not provided on the first or second questionnaire). A person was considered a former smoker if he or she reported having quit smoking at least a year before responding to the most recent questionnaire. A person was considered a current smoker from the date at which he or she first provided information on smoking behaviors on the first or second questionnaire until either the date of reported quitting or the end of follow-up, whichever occurred first. If a smoker did not report quitting on the first or second questionnaire, it was presumed that the person did not quit smoking. Smoking status was considered unknown until the date at which smoking information was first provided.

For participants with a known smoking status, additional time-dependent smoking variables were evaluated, including cigarettes smoked per day, smoking duration, pack-years smoked, and years since quitting smoking. Pack-years smoked was defined as the product of packs smoked per day (20 cigarettes per pack) and number of years smoked as reported on the most recent of the first or second questionnaire. Cigarettes smoked per day was also ascertained from the most recent of the first or second questionnaire. Years since quitting smoking was defined as the difference between attained age and age of quitting smoking, and smoking duration was defined as the difference between attained age and age started smoking for current smokers and age of quitting smoking and age started smoking for former smokers. Age started smoking was defined as the starting age of smoking reported on the most recent questionnaire completed (first or second). These smoking variables were also considered to be unknown prior to the date of the questionnaire that provided the relevant information (or if missing on both the first and second questionnaire). Smoking status, pack-years smoked, and cigarettes smoked per day were evaluated both as confounders and as potential effect modifiers.

Demographic and work history characteristics

In addition to time-dependent smoking characteristics and attained age, we also considered the following fixed variables that were collected on the first or second questionnaire as potential effect modifiers; year of birth, sex, year first worked, and history of working with fluoroscopically-guided interventional (FGI) or nuclear medicine (NM) procedures. FGI and NM procedure work histories were evaluated as potential effect modifiers as these technologies are associated with higher annual badge doses.

Statistical analyses

The excess relative risk (ERR=RR - 1) is the proportional increase in risk due to exposure over the background risk (in the absence of exposure). In this study, we have assumed that the ERR is a linear function of dose (or ERR = β d, where d is dose). ERRs per 100 mGy of lung absorbed dose and 95% confidence intervals (CIs) were calculated using Poisson regression models. To account for the latency period between exposure and lung cancer onset, cumulative absorbed lung dose was treated as a time-dependent variable and lagged 10 years, a lag commonly used in other occupational studies of lung cancer in radiation workers.^8,10,21 For these analyses, a person-year table was created that stratified events and person-years by attained age (5-year intervals), birth cohort (5-year intervals from 1930-1960), sex, year first worked (before 1950, 1950-1959, 1960-1969, 1970 or later, unknown), ever worked with FGI procedures (no, yes, unknown), ever worked with NM procedures (no, yes, unknown), smoking status (never, former, current, unknown), pack-years smoked (never smoker, 1-<20, 20-<30, 30-<40, ≥40, unknown), cigarettes smoked per day (never, ≤10, >10, unknown), years since quit smoking (never smoker, current smoker, <20, 20-30, ≥30, unknown), and cumulative lung absorbed dose (10 mGy categories). All models were stratified on attained age, sex, and birth cohort and adjusted for pack-years smoked and years since quit smoking using categorical variables. Smoking duration and age started smoking were also considered, but not included in the final analysis as smoking duration was highly correlated with pack-years smoked and age started smoking did not significantly improve the model fit.

Analyses were based on a linear excess relative risk (ERR) model in which the risk relative to the baseline (defined by attained age, birth cohort and sex) is given by B [1 + β d] where B is a log-linear function of pack-years and years since quitting smoking, d is cumulative lung absorbed dose, and β is the ERR expressed per 100 mGy. We used likelihood ratio methods to test the null hypothesis that the β equals 0 and to evaluate confidence intervals. The likelihood ratio test was also used to compare the deviance of models with separate βs by categories of the modifying variables to the deviance from a model with a single overall β.

The Epicure AMFIT module was used to carry out the analyses.²² All statistical tests were two-sided, and p-values less than 0.05 were considered statistically significant.

Data availability

The data that support the findings of this study are available on request from the corresponding author.

RESULTS

During follow up, 1,090 of 106,068 participants died from lung cancer (379 men and 711 women). The mean follow-up time from completion of the baseline questionnaire to death, end of the study, or loss to follow-up was 24 years (maximum 29 years). Cohort participants were primarily women (76%) and born in the 1940s or later (81%). Demographic, lifestyle, and work history characteristics stratified by case status are shown in Table 1. Cases were more likely to have smoked for a greater duration and at a higher intensity than non-cases. Cases were also more likely to have started working as a radiologic technologist in earlier decades than non-cases.

Table 1.

Demographic, lifestyle, and work history characteristics of cases and non-cases, United States Radiologic Technologists Study, 1983-2012, (n = 106,068)

Characteristic	Cases	Non-cases
	N (%)	N (%)
Total	1,090 (100)	104,978 (100)
Age at baseline (years) ≤40 40-<50 50-<60 ≥60	140 (13) 320 (29) 346 (32) 284 (26)	57,240 (55) 27,999 (27) 12,457 (12) 7,282 (7)
Year of birthBefore 1930 1930-39 1940 or after	348 (32) 357 (33) 385 (35)	7,753 (7) 12,778 (12) 84,447 (80)
Sex Men Women	379 (35) 711 (65)	25,509 (24) 79,469 (76)
Year first worked as a radiologic technologist Before 1950 1950-1959 1960 or later Unknown	222 (20) 372 (34) 442 (41) 54 (5)	4,839 (5) 12,230 (12) 83,443 (79) 4,466 (4)
Ever worked with FGI procedures No Yes Unknown	461 (42) 147 (13) 482 (44)	56,027 (53) 21,343 (20) 27,608 (26)
Ever worked with NM procedures No Yes Unknown	393 (36) 190 (17) 507 (47)	54,670 (52) 21,022 (20) 29,286 (22)
Smoking status Never smoked Ever smoked Unknown	89 (8) 972 (89) 29 (3)	50,223 (48) 53,678 (51) 1,077 (1)
Pack-years smoked Never smoked <20 20-<30 ≥30 Unknown	89 (8) 188 (17) 163 (15) 602 (55) 48 (4)	50,223 (48) 35,543 (34) 7,474 (7) 9,475 (9) 2,263 (2)
Cigarettes smoked per day Never smoked ≤10 >10 Unknown	89 (8) 121 (11) 860 (79) 20 (2)	50,223 (48) 20,572 (20) 32,903 (31) 1,280 (1)

Participants reported working an average of 17 years as a radiologic technologist (maximum 62 years). The mean cumulative lung absorbed dose from occupational radiation exposure at the end of follow-up was 25 mGy (range 0-810 mGy). These doses were higher and more variable for participants who began working in earlier compared to more recent decades (Table 2).

Table 2.

Dose and work history characteristics by decade first worked, United States Radiologic Technologists Study, 1983-2012, (n = 106,068)

Decade first worked	Total participants	% Women	Mean total years worked (+/− SD)	Mean (range) cumulative lung dose in mGya
Before 1940	853	75	25±14	200 (0-810)
1940-1949	4,208	65	22±13	78 (0-260)
1950-1959	12,602	74	20±12	30 (0-170)
1960-1969	32,002	79	19±10	17 (0-100)
1970-1979	50,177	76	16±8	9.3 (0-78)
1980 or later	1,706	72	10±6	3.4 (0-77)
Unknown	4,520	66	11±11	8.4 (0-76)
Total across all decades	106,068	76	17±10	25 (0-810)

Abbreviations: mGy – milliGray, ARRT – American Registry of Radiologic Technologists

^aMean cumulative lung absorbed dose from occupational radiation exposure at the end of follow-up (10-year lag applied)

A positive, but non-significant, dose-response relationship of radiation exposure with lung cancer mortality was observed after adjusting for attained age, birth cohort, and sex (ERR per 100 mGy: 0.16, 95% CI: <0-0.46; p-trend=0.12) (Table 3). However, the ERR was attenuated after further adjustment for pack-years smoked and years since quit smoking (ERR per 100 mGy: −0.02, 95% CI: <0-0.13; p-trend>0.50). Similar results were observed in models using a 0-, 5-, and 15- year exposure lag (data not shown). Inclusion of employment duration in the model did not greatly change the overall effect estimate (ERR per 100 mGy: 0.07, 95% CI: <0-0.34; p-trend=0.44).

Table 3.

Dose-response estimates for lung cancer mortality risk, overall and by selected factors, United States Radiologic Technologists Study, 1983-2012, (n = 106,068)

Characteristic	Deaths	PYR	Mean (range) cumulative lung dose in mGya	ERR per 100 mGy (95% CI)b	p-trend for dose
Overall	1,090	2,526,265	25 (0-810)	−0.02 (<0-0.13)	>0.50
Overall (unadjusted for smoking)c	1,090	2,526,265	25 (0-810)	0.16 (<0-0.46)	0.11
Demographic Characteristics
Attained age ≤60 60-<70 ≥70	260 358 472	1,951,614 395,179 179,472	14 (0-180) 27 (0-360) 53 (0-810)	0.75 (<0-3.55) 0.14 (<0-1.03) −0.03 (<0-0.15) p-het > 0.50	0.40 >0.50 >0.50
Year of birth Before 1930 1930-39 1940 or after	348 357 385	142,980 300,892 2,082,393	67 (0-810) 25 (0-170) 12 (0-99)	−0.05 (<0-0.09) 1.13 (0.27-2.42) 0.49 (<0-2.13) p-het = 0.01	0.39 0.01 0.46
Sexd Men Women	379 711	579,228 1,947,036	26 (0-670) 24 (0-810)	−0.14 (<0-0.09) 0.06 (<0-0.23) p-het = 0.18	0.24 0.49
Work History Characteristics
Year first worked as a radiologic technologist Before 1950 1950-1959 1960 or later	222 372 442	88,140 278,442 2,078,187	94 (0-810) 30 (0-170) 13 (0-100)	−0.06 (<0-0.14) 1.31 (0.20-3.50) −0.27 (<0-1.20) p-het = 0.05	0.34 0.01 >0.50
Ever worked with FGI procedures No Yes	461 147	1,366,378 518,053	24 (0-670) 22 (0-750)	−0.01 (<0-0.25) 0.22 (<0-0.51) p-het = 0.25	>0.50 0.24
Ever worked with NM procedures No Yes	393 190	1,335,451 507,834	23 (0-670) 24 (0-750)	−0.17 (<0-0.15) 0.27 (<0-0.54) p-het = 0.11	0.28 0.12
Smoking History Characteristics
Smoking status Never smoked Ever smoked	89 972	1,213,054 1,289,165	33 (0-810) 24 (0-750)	0.37 (<0-1.50) −0.05 (<0-0.12) p-het = 0.27	0.18 >0.50
Pack-years smoked Never smoked <20 20-<30 ≥30	89 188 163 602	1,213,054 876,778 177,426 211,764	33 (0-810) 22 (0-690) 23 (0-590) 29 (0-750)	0.44 (<0-1.60) 0.67 (0.05-1.78) 0.13 (<0-0.86) −0.04 (<0-0.13) p-het = 0.06	0.14 0.03 >0.50 >0.50
Cigarettes smoked per day Never smoked ≤10 >10	89 121 860	1,213,054 507,637 782,916	33 (0-810) 24 (0-670) 24 (0-750)	0.38 (<0-1.50) 0.66 (<0-1.92) −0.05 (<0-0.11) p-het = 0.04	0.17 0.05 0.46

Abbreviations: PYR – person-years at risk; ERR – excess relative risk; mGy – milliGray; CI – confidence interval, FGI – fluoroscopically-guided interventional, NM – nuclear medicine; p-het – p-value for test of heterogeneity across categories of a given characteristic

^aMean cumulative lung absorbed dose from occupational radiation exposure at the end of follow-up (10-year lag applied)

^bAdjusted for attained age, sex, year of birth, pack-years smoked, and years since quit smoking

^cAdjusted for attained age, sex, and year of birth

^dModeled on the log-linear scale to facilitate model convergence

Statistically significant heterogeneity was observed by birth-year (p=0.01) with a statistically significant positive dose-response observed for those born in the 1930s (ERR per 100 mGy: 1.13, 95% CI: 0.27-2.42; p-trend=0.01), and a negative dose-response for those born before 1930. Similarly, we observed a statistically significant positive association for those who first worked in the 1950s (ERR per 100 mGy: 1.31, 95% CI: 0.20-3.50; p-trend=0.01) and a negative dose-response for those who first worked before 1950. Birth-year and year first worked as a radiologic technologist were moderately correlated (r=0.69). The risk did not vary significantly by attained age or sex (p-heterogeneity >0.50 and 0.18, respectively). The results also did not differ for those who ever (versus never) worked with FGI or nuclear medicine procedures (p-heterogeneity 0.25 and 0.11, respectively).

The magnitude of the dose-response relationship was greater in never smokers (ERR per 100 mGy: 0.37, 95% CI: <0-1.50) than ever smokers (ERR per 100 mGy: −0.05, 95% CI: <0-0.12), but this difference was not statistically significant (p-heterogeneity=0.27) (Table 2). There was evidence of effect modification by pack-years smoked (p-heterogeneity=0.06), and cigarettes smoked per day (p-heterogeneity=0.04) among smokers. There was a significant dose-response for non- and light smokers who smoked 0-<20 pack-years combined (ERR per 100 mGy: 0.41, 95% CI: 0.01-1.15; p-trend=0.04), but not for heavier smokers who smoked ≥20 pack-years (ERR per 100 mGy: −0.03, 95% CI: <0-0.15; p-trend>0.50) (Figure 1). A borderline significant dose-response was also observed for those who reported smoking ≤10 cigarettes per day (ERR per 100 mGy: 0.66, 95% CI: <0-1.92; p-trend=0.05), but no association was observed among those smoking >10 cigarettes per day.

Figure 1. Dose-response for cumulative lung absorbed dose from occupational radiation exposure and lung cancer mortality risk, United States Radiologic Technologists Study, (n = 106,068).

Shown are relative risks and 95% confidence intervals for lung cancer mortality risk (1983 - 2012) plotted at the mean cumulative lung absorbed dose from occupational radiation exposure at the end of follow-up for each dose category. Cumulative lung absorbed dose from occupational radiation exposure from 1916 - 1997 was categorized as < 10 mGy (referent), 10 - < 30 mGy, 30 - < 50 mGy, 50 - < 70 mGy, and ≥ 70 mGy. Relative risks and excess relative risks are adjusted for attained age, sex, birth cohort, and years since quit smoking. (A) shows the dose-response relationship of cumulative lung absorbed dose with lung cancer mortality in non-smokers and light smokers who smoked 0 - < 20 pack-years and (B) shows the dose-response relationship of cumulative lung absorbed dose with lung cancer mortality in heavier smokers who smoked ≥ 20 pack-years.

DISCUSSION

This large cohort of U.S. radiologic technologists provided a unique opportunity to quantify the relationship between low-dose, protracted occupational radiation exposure and lung cancer mortality using high-quality estimates of annual lung-absorbed occupational radiation dose and accounting for the potential confounding effects of cigarette smoking. Although we observed little evidence of a dose-response relationship in the full cohort, statistically significant positive associations were observed for certain subgroups of participants with significant heterogeneity in the dose-response by both year of birth and year first worked. These associations should be interpreted cautiously given that many statistical tests were performed. Heterogeneity by smoking variables was suggested with a statistically significant dose-response for smokers who smoked less than 20 pack-years.

We do not fully understand the underlying causes for the heterogeneity in the ERR by birth cohort and year first worked, particularly the absence of a dose-response for those born before 1930 (or hired before 1950), time periods when doses were highest. For later decades, cumulative lung-absorbed doses are lower on average and so there is limited statistical power to detect a dose-response relationship if one exists. Since there is greater uncertainty in the annual dose estimates before 1960, the dose-response for the earliest decades may have been attenuated. Additionally, since the first questionnaire was not administered until 1983, it is possible that those who received the highest occupational doses, and thus most susceptible to radiation-induced lung cancer, may have died before the questionnaires were administered, which may partially explain the lack of association. Since information on annual occupational radiation doses was not available after 1997, a lag period ranging from 10-15 years was applied to participants who exited the study after 1997. However, we would not expect a substantial underestimation of the worker’s cumulative doses as occupational radiation doses have been decreasing over time.²³ Our results are consistent with an earlier analysis of the work history data from the USRT cohort, in which a greater number of years worked in the 1950s (but not other decades) was associated with increased risk of lung cancer death.¹²

Comparable with findings from the Life Span Study of Japanese atomic bomb survivors, we observed excess radiation-related lung cancer risk among non- and light, but not heavy, smokers. The ERR per 100 mGy was elevated for those who smoked up to 20 pack-years and for those who smoked up to 10 cigarettes per day. In studies of the Japanese atomic bomb survivors; the ERR/Gy for lung cancer increased with up to 10 cigarettes per day, with little indication of excess risk in heavy smokers.^14,15 Sub-multiplicative interaction with cigarette smoking has also been reported in studies of radiation workers exposed to radon and plutonium, where the ERR is greater for non-smokers compared to ever smokers.^7,24 This may be because of the strong dose-response effect of cigarette smoking on lung cancer risk, which would make it difficult to detect a weak radiation effect in heavy smokers.

Our current findings are also compatible with results from most other cohorts exposed to low-dose, protracted radiation, which have tended to provide positive risk estimates with confidence intervals that include zero (Table 4). The International Nuclear Workers Study (INWORKS) cohort study of 308,297 nuclear workers provided evidence of a weak dose-response association with lung cancer; however, the results were unadjusted for smoking status, which may act as a confounder (ERR per 100 mGy: 0.05, 90% CI: 0.00-0.11).¹⁰ In our study, we observed a slightly stronger (but non-significant) positive association in models that were unadjusted for smoking history. Lung cancer has also been linked with exposure to inhaled alpha particles from radon and plutonium in miners and nuclear workers.^7,25

Table 4.

ERR estimates for lung cancer incidence and mortality risk in select cohorts exposed to external low-LET radiation

First author, year	Study population	Follow-up period	Type of radiation exposurea	Cases	Effect measure	Adjusted for smoking?
Velazquez-Kronen, 2018 (current study)	106,068 US radiologic technologists (25,888 men and 80,180 women)	1983-2012 (Mortality)	x-ray	1,090	ERR per 100 mGy: −0.02 95% CI: <0-0.13	Pack-years smoked (Never smoker, 1-<20, 20-<30, 30-<40, ≥40, unknown), years since quit smoking (never smoker, current smoker, <20, 20–30, ≥30, unknown)
Richardson, 2018	308,297 nuclear workers in France, UK, and US (268,262 men and 40,035 women)	1944-2005 (Mortality)	Gamma, x-ray, neutron, beta particles	5,802	ERR per 100 mGy: 0.05 90% CI: 0-0.11	No
Cahoon, 2017b	105,444 Japanese atomic bomb survivors (42,910 men and 62,534 women)	1958-2009 (Incidence)	Gamma and neutron	2,741	ERR per 100 mGy: 0.12 95% CI: 0.07-0.18	Smoking status (never, past, current, unknown), average number of cigarettes per day, years smoked and years since quitting
Davis, 2015	17,435 Techa River residents (7,521 men and 9,914 women)	1956-2007 (Incidence)	Gamma, internal 137Cs, 90Sr, 89 Sr and other uranium fission products	339	ERR per 100 mGy: 0.01 95% CI: <0-0.18	Smoking status (ever smoker, male non-smoker)
Gilbert, 2013	14,621 Mayak nuclear facility workers (10,918 men and 3,703 women)	1948-2008 (Mortality)	Gamma	486	ERR per 100 mGy: 0.01 95% CI: <0-0.04	Sex-specific smoking status (non-smoker, smoker, unknown)
Muirhead, 2009c	174,541 UK nuclear workers (157,505 men and 17,036 women)	1955-2001 (Mortality)	Gamma, x-ray, neutron, beta particles	2,433	ERR per 100 mGy: 0.01 95% CI: <0-0.08	No

Abbreviations: USRT – US radiologic technologists; mGy – milliGray; ERR – excess relative risk; CI – confidence interval

^aTypes of radiation exposures accounted for in the dose estimation

^bERR for female non-smokers (at age 70 after radiation exposure at age 30)

^cA subset of the International Nuclear Workers Study (INWORKS) study included in Richardson et al., 2018

A strength of this study is the availability of detailed smoking information which allowed us to assess the confounding and modifying effects of smoking status to better understand the association of radiation exposure with lung cancer. Moreover, this cohort is comprised predominantly of women, who have been underrepresented in other cohorts. Additional strengths of our current study include a comprehensive dose reconstruction, large numbers of lung cancer deaths and long-term follow-up in a large population of radiologic technologists. This study also has several limitations. The use of self-reported information on smoking behaviors, which was not continuously updated over follow-up and partially missing, could result in residual confounding. While an assessment of dose-response among never smokers would be least susceptible to confounding, the statistical power to detect an association if one exists in this subgroup was limited by the small number of cases (n = 89); in this group, we saw evidence of a weak trend that was not significant. We were also unable to evaluate heterogeneity in lung cancer risk by histologic types.

In summary, we did not observe an association between cumulative lung dose and lung cancer mortality in the overall USRT cohort. There appeared to be a dose-response relationship among subsets of technologists who were born in the 1930s and first started working in the 1950s. There was also some evidence that the radiation-related risk was greater for non- and lighter cigarette smokers than for heavier smokers. These findings should be evaluated further in pooled studies with other radiation-exposed cohorts with detailed information on lung dose and smoking behaviors, to better understand the modifying effects of smoking on radiation carcinogenesis and to quantify more precisely the relationship between radiation and lung cancer in non- and light smokers.

Novelty and Impact:

Exposure to low-dose, protracted radiation has been associated with increased lung cancer risk; however, few studies have examined the confounding and modifying effects of cigarette smoking. In a cohort of radiologic technologists, we found no overall evidence of excess lung cancer in relation to occupation radiation dose, but some evidence of increased risk for light compared to heavier smokers. This indicates sub-multiplicative interaction between low-dose radiation and smoking, which warrants further evaluation in pooled studies.

Abbreviations:

ARRT

American Registry of Radiologic Technologists

CI

confidence interval

ERR

excess relative risk

FGI

fluoroscopically-guided interventional

mGy

Milligray

NM

nuclear medicine

RR

relative risk

USRT

United States Radiologic Technologists