EPA Radiogenic Cancer Risk Projections for the U.S. Population Michael Boyd Radiation Protection Division U.S. Environmental Protection Agency 2011 OAS Annual Meeting Richmond, VA August 24, 2011 EPA’s Risk Estimates • Old “Blue Book” published (1994) • Addendum on uncertainties with some revisions (1999) • FGR-13: radionuclide specific estimates (1999) 2 NAS BEIR VII Rept. (2006) • Co-sponsored by EPA and several other Federal agencies • Developed models for estimating low dose, low-LET radiogenic cancer risk as a function of: – – – – – • Dose Cancer site Sex Age-at-exposure Age-at-diagnosis (death) Developed quantitative uncertainty distributions 3 BEIR VII Treatment of Low Dose Issue Endorsed LNT: “…the balance of evidence from epidemiologic, animal and mechanistic studies tend to favor a simple proportionate relationship at low doses between radiation dose and cancer risk.” Derived a DDREF of 1.5 for solid cancers based on a “Bayesian” analysis of epidemiological and radiobiological data 4 BEIR VII Risk Models • Most models derived from Japanese Life Span Study (LSS) incidence data • Breast and thyroid risks from pooled analyses of medical and LSS cohorts • Most solid cancer risk models were assumed to have the same age and temporal dependency • Two models for most solid cancers, EAR and ERR: – The excess absolute risk (EAR) adds to the baseline cancer incidence or mortality rate – The excess relative risk (ERR) multiplies the baseline cancer rate 5 Cancer Risk Coefficients • Excess absolute risk (EAR): rate of radiogenic cancers per unit dose • Excess relative risk (ERR): fractional increase in cancer rate per unit dose; i.e., radiogenic cancer rate per unit dose divided by baseline rate in the population 6 EPA Modifications and Additions to BEIR VII Approach • Additional cancer sites (bone, kidney, skin) • High-LET (α-particle) risk estimates • Stationary population • Weighted arithmetic mean of ERR & EAR estimates rather than geometric mean • Breast cancer analysis • Risk estimate for prenatal exposures • Expanded uncertainty analysis 7 Risk (Gy-1) from Uniform, Whole-Body Irradiation INCIDENCE 2011 1999 MORTALITY 2011 1999 MALES 9.6x10-2 6.5x10-2 4.7x10-2 4.6x10-2 FEMALES 1.4x10-1 1.0x10-1 6.9x10-2 6.8x10-2 ALL 1.2x10-1 8.5x10-2 5.8x10-2 5.7x10-2 8 Age-Dependence of Risk from Whole-Body Irradiation LAR per Gy 0.4 0.3 Incidence —— Mortality – – – – Incidence ——— Mortality ——— 0.2 0.1 0 0 20 40 60 80 Age at exposure 9 Life-Table Calculations Age- and Sex-Specific Survival Functions • • • • • Life-table calculation: corrects for competing causes of death Survival function, S(a,ae): probability of surviving to age a, conditioned on being alive at age of exposure, ae r(a,ae) = f(D) g[a,ae,r0(a)] S(a,ae) Lifetime Attributable Risk due to an exposure at age ae: LAR (ae) = ∑ r(a,ae) Population risk: ∑ N(ae) LAR (ae) 10 Stationary Population • Age distribution constant – # births = # deaths, each year – N(a) = N0 S(0,a) • Risk per Gy for (small) acute exposure = that for chronic low dose rate exposure • Stationary population “older” than the U.S. population ===> estimated radiogenic cancer risk is lower 11 Averaging Model Projections • Weighted Arithmetic Mean AM = w(EAR) + (1-w)(ERR) • Weighted Geometric Mean log GM = w log (EAR) + (1-w) log (ERR) • GM ≤ AM • For most sites, BEIR recommended GM and w = 0.3, GM = (EAR).3 (ERR).7 Exceptions – Lung (wts. reversed), Thyroid, Breast 12 LAR Projections for Incidence (x10-4 Gy-1) Cancer Site Males Females Sex-Averaged Stomach 62 75 68 Colon 146 92 119 Lung 130 308 220 Breast — 289 146 Leukemia 92 69 80 Other 525 518 522 (Skin) (182) (96) (138) Total 955 1351 1155 13 Other Revisions & Extensions to BEIR VII • Thyroid: Model primarily based on NCRP Thyroid Report (NCRP No. 159) • Skin: Model projection for BCC but not included in total incidence estimate • Alpha-particles – For most sites, RBE=20 – For leukemia, RBE=2 • Bone cancer – Model based on studies of patients injected with 224Ra – Low-LET estimate derived assuming that RBE=20 14 Major Sources of Uncertainty • Sampling errors • Transport of risk from the LSS cohort to the U.S. population • DDREF • Age/temporal dependence 15 Uncertainty in Low-Dose “Extrapolation” BEIR VII/EPA implicitly assumes: (1) LQ dose-response: R = α D + β D2 (2) β/ α, DDREF same for all solid cancers 95% CI on DDREF then turns out to be ≈ a factor of 2, up or down 16 Uncertainties Projection (per 104 person-Gy) Uncertainty Interval (90%) Stomach 68 (9, 220) Colon 119 (42, 220) Liver 30 (6, 94) Lung 220 (83, 420) Breast 146 (70, 290) Prostate 44 (0, 200) Thyroid 44 (15, 140) Leukemia 80 (29, 160) 1180 (560, 2130) Cancer Site Total 17 Other Data Pertaining to Low Dose Risks I. Radiobiology • Experiments indicate novel phenomena not easily reconciled with standard paradigm for radiation carcinogenesis – Adaptive response, bystander effect, genomic instability, etc. – Mostly based on cellular studies, inconsistency among laboratories – significance for human carcinogenesis unclear – May produce increased as well as decreased risks 18 Other Data Pertaining to Low Dose Risks II. Epidemiology • Epidemiological data on cohorts receiving fractionated or chronic exposures show excess risks, generally consistent with a DDREF of about 1. – Fractionated: Multiple fluoroscopies (breast cancer) Scoliosis patients (breast cancer) Tinea capitis patients (thyroid) – Chronic: British nuclear workers (leukemia, solid cancers) Chernobyl liquidators (leukemia) Techa R. Cohort (leukemia, solid cancers) Taiwanese apartment dwellers (leukemia) Semipalatinsk test-site residents (solid cancers) 19 Future Plans • New “Blue Book”: Available on RPD web-site at www.epa.gov/radiation • Addendum on RBE for low energy electrons and photons to be published by ORNL – Important for 3H betas, medical X-rays, & perhaps certain external radionuclide exposure situations – NCRP committee also working on such a report • New Federal Guidance Report on risks from external and internal radionuclide exposures – Replace FGR-13 – Use new risk models in combination with latest ICRP dosimetry 20