EFFECT OF AGE AT EXPOSURE IN 11 UNDERGROUNDMINERS STUDIESL. Tomasek*National Radiation Protection Institute, Prague, Czech Republic

*Corresponding author: ladislav.tomasek@suro.cz

Eleven underground miners studies evaluated the risk of lung cancer from exposure in underground mines. Nearly 68 000 minerswere included in the joint study, contributing to nearly 2700 lung cancers. The resulting model of the Biological Effects ofIonizing Radiation (BEIR) VI Committee considered linear exposure response relationship, which was modified by time since ex-posure (TE), attained age and exposure rate. The effect of age at exposure (AE) was not explicitly evaluated. The presentationaims to show that the modifying effect of AE is substantial if time-since-exposure modification is simultaneously used in themodel. When the excess relative risk per unit exposure (ERR/WLM) is adjusted for TE, the ERR/WLM corresponding to AE<15 is 0.013 and in subsequent categories decreased gradually up to the AE of 40 and more years, which was only 0.004. In com-parison with the BEIR VI model, the present model predicts higher risks at younger ages and the risk decreases more rapidly.

INTRODUCTION

Cohort studies conducted among uranium minershave made a substantial contribution to understand-ing the risk of radon. Many studies have already beenpublished separately and jointly. One such analysis of11 cohorts coordinated by the National CancerInstitute (NCI) provided the most comprehensiveresults(1, 2). The present analyses are based on datafrom 11 studies reported by the Biological Effects ofIonizing Radiation (BEIR) VI committee.

The aim of the present study is to analyse the effectof age at exposure. This effect was not explicitly evalu-ated in the BEIR VI report. Some analyses in thissense were done in the original report of these cohortsconducted by the NCI(2). The estimates of excess rela-tive risk per WLM (ERR/WLM) for separate studiesand the age at first exposure categories from thisreport are shown in Figure 1. The effect of age at firstexposure was significant only in the China study.In the Czech study, the significant trend was in theopposite direction.

METHODS

Exposure estimates

The exposure estimation in the studies relied on mea-surements made for regulatory and research purposes.These measurements were extended using interpol-ation and extrapolation to complete gaps for mines inparticular years. Additionally, missing informationfor mines in the earliest years of some of the studieswas completed by either expert judgement or by re-creation of operating conditions. In addition, in theearly years of mining, measurements of radon ratherthan radon progeny were made. The numbers of mea-surements made also varied widely across the different

studies included in the pooled analysis and withinstudies the numbers of measurements tend to begreater in the later years of operation of the mines,when the exposures were generally lowest(1). In somestudies exposure to radon was accompanied by expos-ure to other carcinogens like arsenic (Ontario, China).Some miners also worked at other non-uraniummines and these exposures were not always covered intheir working histories.

Statistical methods

Relative-risk regression procedures were applied todata summarised in a multi-way table, consisting ofnumbers of lung cancer, person-years and summaryvariables for each cell of the cross-tabulation. Ana-lyses were conducted using the EPICURE software(3).Data were cross-classified by various factors, such as

Figure 1. ERR/WLM in 11 studies by age at first exposure.

# The Author 2014. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Radiation Protection Dosimetry (2014), pp. 1–4 doi:10.1093/rpd/ncu068

Radiation Protection Dosimetry Advance Access published April 20, 2014 at U

niversidade de BrasÃ

­lia on May 4, 2014

http://rpd.oxfordjournals.org/D

ownloaded from

attained age, calendar period or cumulated exposure.Similarly as in the BEIR IV and VI reports(1, 4), cumu-lated exposure to radon progeny was divided into severalexposure windows defined by time or other potentialmodifying factors such as the exposure rate. The cross-classification was based on cumulated exposure in suchwindows.

Some modifying factors were based on approachesreported in refs (5, 6). The modifying effects of TE andAE were accounted for by exponential terms in themodel as follows:

ERR ¼X

bCWC

� �expðcAðAE� 30Þ

þ cTðTE� 20ÞÞ; ð1Þ

where WC are cumulated exposures from exposures forwhich the annual exposure rates are in the respectivecategory C (,4 WL, 4–8 WL and .8 WL), AE is ageat median exposure and TE is time since median expos-ure. These two variables are calculated with referenceto median exposure i.e. when half of current cumulatedexposure was reached. Age at median exposure is theage corresponding to this value. This value is amedium between the age at first exposure and age atlast exposure. The corresponding time since median ex-posure (TE) is related to age at median exposure (AE)and attained age (AA) by

AEþ TE ¼ AA: ð2Þ

Similarly as in the BEIRVI report(1), all estimates con-sidered in this study are stratified by cohort, attainedage and arsenic exposure.

RESULTS

The results are based on joint cohorts representingnearly 68 000 miners with 2852 lung cancer cases. Thecohorts are different in their size, duration of follow-up and particularly in exposures levels reflecting dif-ferent periods of mining. In some studies, radon ex-posure is combined with exposure to arsenic. Thesummary of cohorts considered in this study is givenin Table 1.

The present analyses aim at evaluating the agemodification of the risk. Most of lung cancers in the

Table 1. Summary description of 11 studies of underground miners.

Study Ore Miners Follow-up Lung cancers Exposure Measured since WLM

USAColorado U 3346 1950–90 377 1936–68 1949 606New Mexico U 3286 1957–85 57 1953–85 1950s 87

CanadaOntario U þAu 21 346 1955–86 291 1937–85 1955 24Beaverlodge U 8486 1950–80 65 1950–80 1954 12Port Radium U 2103 1950–80 57 1942–60 1945 138Newfoundland CaF2 2092 1950–90 147 1936–77 1960 252

EuropeWest Bohemia U 4320 1952–90 705 1948–65 1949 189France U 1785 1948–85 45 1946–85 1953 53Malberget Fe 1294 1951–91 79 1908–77 1968 77

Asia and AustraliaRadium Hill U 2516 1948–87 54 1948–61 1954 4China Sn 17 120 1976–87 975 1922–76 1990 249

Total 67 694 2852 133

Table 2. Summary by attained age.

Age PY Lung cancers SMR

,40 473 876 45 4.1540–49 329 379 381 4.2050–59 243 999 1101 4.2560–69 97 511 1016 3.9970þ 29 378 309 3.16

Table 3. Summary by AE.

AE PY Lung cancers SMR

,15 25 155 103 7.7615–19 94 108 192 8.5020–24 272 494 315 6.1825–29 265 803 451 5.1330–34 195 167 455 4.1335–39 132 301 441 3.8140þ 186 165 895 2.87

L. TOMASEK

Page 2 of 4

at Universidade de B

rasília on M

ay 4, 2014http://rpd.oxfordjournals.org/

Dow

nloaded from

joint study are in the age category of over 50 y.However, the cases observed below this age are of im-portance in analysing the age-modifying effects of therisk. The distribution of person-years, observed lungcancers and standardised mortality rate (SMR) byattained age is given in Table 2. The SMRs show amoderate decreasing trend with attained age. The de-creasing trend of SMR with AE is even stronger(Table 3).

The crude excess relative risks per WLM (ERR/WLM) in categories of AE are not significantly differ-ent (Table 4). However, when the ERR/WLMs are

adjusted to TE (ERR/WLM corresponding to ex-posure window 5–14 y), the effect of AE is significant( p ¼ 0.004).

Trends of ERR/WLM in combined categories ofAE and TE are given in Table 5 together withnumbers of lung cancer. Although the marginal trendby AE (last line of Table 5) is indistinctive, thesetrends for each category of TE are distinct (with theexception of first category of AE and two categoriesof TE, which are based on two and six cases).Simultaneously, the trends with TE in categories ofAE are more pronounced.

Table 4. Estimates of ERR/WLM by categories of AE.

AE ERR/WLMa 95 % CI ERR/WLMb 95 % CI

,15 0.0031 0.0017–0.0054 0.0131 0.0066–0.026015–19 0.0029 0.0020–0.0042 0.0108 0.0066–0.017620–24 0.0032 0.0024–0.0043 0.0109 0.0074–0.016125–29 0.0034 0.0027–0.0043 0.0097 0.0070–0.013630–34 0.0033 0.0025–0.0042 0.0075 0.0055–0.010235–39 0.0039 0.0031–0.0050 0.0069 0.0052–0.009340þ 0.0038 0.0030–0.0048 0.0044 0.0034–0.0057

p ¼ 0.623 p ¼ 0.0004

aCrude estimate stratified by age, study and arsenic exposure.bEstimate adjusted for TE windows and exposure rate ,4WL and stratified by age, study and arsenic exposure.

Table 5. Numbers of lung cancer and estimates of ERR/WLMa by categories of AE and TE.

AE

TE ,20 20–29 30–39 40þ Overall

,15 2 0.0115 49 0.0221 162 0.0147 407 0.0096 620 0.011815–24 6 0.0067 198 0.0102 382 0.0081 315 0.0086 901 0.009125þ 287 0.0065 519 0.0063 352 0.0070 173 0.0064 1331 0.0070Overall 295 0.0081 766 0.0083 896 0.0094 895 0.0103 2852 0.0091

aAdjusted for exposure rate ,4 WL.

Table 6. Estimates of ERR/WLM by exposure rate (WL) categories with modification by AE and TE, model (1).

ERR/WLM 95 % CI 95 % CI

Exposure rate,4 WL b,4 ¼ 0.0137 0.0102–0.01864–8 WL b4 – 8 ¼ 0.0069 0.0045–0.0101.8 WL b.8 ¼ 0.0049 0.0037–0.0066

Relative change per decade ofAE 0.477 0.377–0.588 ca ¼ 20.074 20.098 to 20.053TE 0.396 0.313–0.487 ct ¼ 20.093 20.012 to 0.072

EFFECT OF AGE AT EXPOSURE IN STUDIES OF MINERS

Page 3 of 4

at Universidade de B

rasília on M

ay 4, 2014http://rpd.oxfordjournals.org/

Dow

nloaded from

The final complete model of the risk that incorpo-rates the modifying effects of exposure rate, TE andAE is given in Table 6. Here, the effect of TEdecreases by 60 % per decade and simultaneously theeffect of AE decreases by 52 % per decade.

In comparison with the resulting model of BEIRVI report, the present model predicts higher risks atyounger ages and the risk decreases more rapidly(Figure 2).

FUNDING

The work was supported by the NCI in 1996 and bythe Czech Ministry of Interior (MV-25972).

REFERENCES

1. National Research Council, Committee on BiologicalEffects of Ionizing Radiation (BEIR VI). Health risks ofexposure to radon. National Academy Press (1999).

2. Lubin, J. et al. Radon and lung cancer risk: a joint analysisof 11 underground miners studies. NIH Publ No 94-3644.National Cancer Institute (1994).

3. Preston, D. L., Lubin, J. H. and Pierce, D. A. EPICURE:risk regression and data analysis software. HiroSoftInternational Corporation (1990).

4. National Research Council, Committee on BiologicalEffects of Ionizing Radiation (BEIR IV). Health risks ofradon and other internally deposited alpha-emitters.National Academy Press (1988).

5. Sevc, J., Tomasek, L., Kunz, E., Placek, V., Chmelevsky,D., Barclay, D. and Kellerer, A. M. A survey of theCzechoslovak follow-up of lung cancer mortality inuranium miners. Health Phys. 64, 355–369 (1993).

6. Tomasek, L., Rogel, A., Tirmarche, M., Mitton, N. andLaurier, D. Lung cancer in French and Czech uraniumminers—risk at low exposure rates and modifying effects oftime since exposure and age at exposure. Radiat. Res. 169,125–137 (2008).

Figure 2. Relative risk from cumulated exposure at the ageof 15–54 y at 6 WLM y21 according to the BEIRVI model(continuous line) and the present model with modifier by

AE and TE (dashed line).

L. TOMASEK

Page 4 of 4

at Universidade de B

rasília on M

ay 4, 2014http://rpd.oxfordjournals.org/

Dow

nloaded from