SCIENTIFIC CORRESPONDENCE been shown to bind to the CCAAT-con­ Secondary structure predictions' for association reflects confounding between taining Y box of HLA class II genes'. these sequences are somewhat incon­ exposure rate and total exposure: workers Conceivably, the conserved sequence clusive but it is interesting that there with lower exposure rates tend to have in all these proteins represents a novel are six invariant glycine residues, sugges­ lower cumulative exposure. But if this DNA recognition motif, one which is ting that critical turns or bends are were the case, one would expect that the found in both prokaryotes and important features of the tertiary slope of the dose-response relationship at eukaryotes. structure. high exposure rates (Colorado, Port GRAEME WISTOW Radium) would be greater at lower than at 1. Goldstein, J., Pollitt, N.S. & Inouye, M. Proc. natn. Acad. Sci. U.S.A. 87, 283-287 (1990). Laboratory of Molecular and higher total dose. Such an effect is indeed 2. Kanehisa, M. IDEAS Users Manual (FCRF, Frederick, Mary• National Eye Institute, apparent in the Colorado data, for which land, 1986). 3. Sakura, H. etal. Gene 73, 499-507 (1988). Developmental Biology, reason the BEIR IV committee used only 4. Didier, D.K. et al. Proc. natn. Acad. Sci. U.S.A. 85,7322- National Institutes of Health, cumulative lagged exposures of less than 7326 (1988). Bethesda, 5. Garnier, J., Osguthorpe, D.J. & Robson, B. J. molec. Bioi. 2,000 WLM in its analysis. Below this 120,97-120 (1978). Maryland 20892, USA level, the data from Colorado and are inconclusive. Whatever the explanation, the practical Radiation and exposure rate consequences for evaluation of the risks of SIR-Substantial evidence has recently First, Hornung and Meinhardt' have exposure to indoor radon are the same. emerged that the coefficients relating the shown an inverse exposure-rate effect Because the exposure rate is low (about risk of lung cancer to exposure to radia­ among miners on the Colorado 0.1 WLM per year in Britain, for tion from radon in mines differ with the Plateau, They conclude that, among example), estimates of risk are more rate of exposure, and that the relative risk miners with equivalent cumulative appropriately based on studies such as increases with decreasing exposure rate. exposure, the risk of lung cancer is greater those at Beaverlodge and Malmberget This may be due to a genuine exposure at lower levels of exposure for longer rather than on all the available data. rate effect or, possibly, to an effect of total periods of time. Moreover, this effect Relative risks thus estimated will be dose, varies with cumulative exposure. For greater than the BEIR IV estimate of 2.5 The first evidence came from studies1.2 miners of similar age and smoking status, per cent per WLM by between 50 and 100 of the workforces at two uranium mines in and with similar cumulative exposure, per cent (the BEIR IV excess relative risk , operated by Hornung and Meinhardt find that a 90 per at Beaverlodge and Malmberget over Ltd1.2. Surveys at each mine allowed cent reduction in exposure rate is associ­ the overall BEIR IV estimate; see table). the lung cancer risk to be related to ated with a 58 per cent increase in risk for If there should be a further increase of individuals' exposure, in working level miners with cumulative exposure in the relative risk with decreasing dose rate months (WLM) , At Beverlodge (Sa­ range 0-834 WLM, but that a similar below about 5 WLM a year, even these skatchewan) and Port Radium (North­ reduction of exposure rate is associated increased estimates may be too low. The west Territories), the increase in relative with only a 10 per cent increase in risk if only study so far to have reported" direct risk with exposure was found to be 3,28 the cumulative exposure is in the range estimates of the effect at indoor levels of per cent per WLM (95 per cent confidence 834-10,000 WLM. exposure to radon suggests an increase in interval 2,08, 4.48) and 0.27 per cent (95 Second, the BEIR IV committee' has relative risk of about 3.4 per cent per per cent confidence interval 0.11, 043), estimated risks for four cohorts of miners WLM, but this estimate is imprecise (95 respectively. (see table, which includes data from Port per cent confidence interval 0.0, 8.0). The difference between the absolute Radium not available to the committee). Estimation of the effect by extrapolation risks is equally striking, being 20.8 per 106 The estimates shown are specific for ages is complicated by the interaction of radon person-years per WLM at Beaverlodge 55-64 and a period 5-14 years after exposure and smoking, which may not (95 per cent confidence interval 11.8, exposure, thus avoiding the complications be multiplicative', but for non-smokers 29.8) and 3,10 at Port Radium (95 per cent of differences of age structure between the an upper limit can be derived from the confidence interval 1.89,4.32), The chief cohorts and different follow-up intervals. observation that their lifetime risk of lung difference between the two mines seems Again, it seems that lower exposure rates cancer of about 0.6 per cent is, with few to be the average exposure rate, estimated are associated with the higher risks. Third, exceptions, remarkably constant in all at 5 WLM a year at Beaverlodge and 109 a similar effect among Czech miners has populations for whom data are available, at Port Radium, Further evidence that been briefly reported'. despite a variation of at least twofold in risk is related to exposure rate has come By themselves, these studies do not the average national exposure to radon from three sources. exclude the possibility that the observed indoors. SARAH C. DARBY RICHARD DOLL RELATIVE RISKS AND AVERAGE EXPOSURE RATES IN FIVE COHORTS OF MINERS EXPOSED TO RADON Imperial Cancer Research Fund, Cancer Epidemiology Unit, Cohort Average exposure rate Relative risk University of Oxford, (WLM per year) (per cent per WLM) Radcliffe Infirmary, Original* BEIR IVt Oxford OX2 6HE, UK Beaverlodge 5 3.3 5.1 Malmberget 5 3.6 3,6 1. Howe, G.R .. Nair, R.C., Newcombe, H.B., Miller, A.B. & 10 (approx,) 1.3 1.8 Abbatt, J.D. J. natn. Cancerlnst. 77, 357-362 (1986). Colorado 124 :j: 0.9§ 2. Howe, G.R. et al. J. natn. Cancer Inst. 79, 1255-1230 (1987). Port Radium 109 0.27 3. Hornung, RW. & Meinhardt, T.J. Health Phys. 52, 417- 430 (1987). * Estimate given by original authors. 4. Committee on the Biological Effects of Ionizing Radiations (BEIR IV) (National Academy Press, Washington, DC, t BEIR IV estimate, based on internal comparison, for age 55-64, 5-14 years after 1988). exposure. 5. Sevc, J., Kunz, E. & Tomasek, L. Health Phys. 54, 27-46 :j: Original paper gives estimates only after allowing for exposure rate effect and are thus not (1988). directly comparable. 6. New Jersey State Department of Health Technical Report• Phase I (1989). § Based on cumulative lagged exposures of less than 2,000 WLM only. 7. Lubin, J.H. YaleJ. bioi. Med. 61,195-214 (1988). 824 NATURE· VOL 344 . 26 APRIL 1990