Projected lifetime cancer risks @ C. E. LAnp ETAL. baseline rate for thyroid cancer among U.S. females (all races) at age 62 is 12.85 per 100,000 per year (NCI 2008). After multiplying by the Hawaii Tumor Registry ASW rate (11.0) for native Hawaiian females and dividing by the corresponding ASW rate (7.33) for U.S. females, the projected baseline rate per 100,000 per year at age 62 for MI women wascalculated as 12.85 X 11.0 X (7.55) | = 19.28. The computation of baseline rates in this paper is based upon a numberof assumptions, the uncertainties of whichare difficult to quantify. However, it can be argued that, when dealing with a past radiation event, the attributable risk, 1.e., the proportion of total cancer cases related to exposure, has more practical significance than the total number of radiation-related cancers. Attributable risk, for example,is the primary basis in the U.S. for evaluating compensation claims for possible radiationrelated cancer (NIH 1985, 2003; Kocher et al. 2008). Projection of attributable risk is also less sensitive to uncertainties in baseline risk. Models for estimation of radiation-related cancer risk BEIR VII linear dose-response models (NRC 2006, 205 The BEIR VII algorithms express cancer-specific ERR as a sex-specific, parametric function linear in radiation dose, with dose coefficients B,, for males and B, for females. Sex-specific dose response was modified by sex-independentfunctions of age at exposure, attained age, and/or time since exposure. These estimates pertain to the population of atomic bomb survivors studied by the Radiation Effects Research Foundation (RERF) in Hiroshima and Nagasaki, Japan (Preston et al. 2007) and, for thyroid cancer, a pooled analysis of incidence data from seven different studies (Ron et al. 1995). The general form of the expression for ERR per unit dose (i.e., per Gy) for specified solid cancers, as derived from the BEIR VII report (NRC 2006, Table 12-2), is ERR Gy|= B X exply X e+ 7X al, (1) where B, y, and » are uncertain parameters estimated from epidemiological data. If exposure age (years) =30, e = 0; otherwise: e = (exposure age — 30) X 10°’, (2) a = In(attained age X 607’). (3) while Tables 12-1, 12-2, and 12-3) for estimating the excess relative risk (ERR) per unit dose are shownin Table 3 for radiation-related leukemia, for cancers of the thyroid gland, stomach, and colon, and for solid cancers other Parameter B is sex-specific (B,, for males and B, for females), whereas parameters y and 7 are sex-neutral. thyroid and non-melanomaskin cancer, were obtained by subtraction. tive fallout. From BEIR VI (NRC 2006, Table 12-3), the than thyroid and non-melanoma skin cancer. Agespecific and lifetime risks for a “residual” category of solid cancers, leaving out stomach and colon as well as For leukemia other than chronic lymphocytic leukemia (CLL), the dose response is a quadratic function of dose for acute exposures but is linear in dose for exposures protracted in time such as those from radioac- Table 3. Values of parameters for BEIR VII dose-response models for excess relative risk: ERR = B X D X expfy Xe +ynXat+6Xt+oXeX th. Here, B (8, for males and 8, for females), y, 7, 6, and @ are uncertain parameters; D is radiation dose in Gy; e = 0 for exposure age =30 and e = (exposure age —30) X 107’ otherwise; a = In(attained age X 607’) and ¢ = In[(attained age — exposure age) X 257]. In the table, point estimates of parameters are given with 95% uncertainty limits in parentheses except where treated in BEIR VII as constants (y and 7 for cancers of the thyroid gland, stomach and colon). The sex-specific parameters B,, and 6, are assumed to have lognormal uncertainty distributions for solid cancers and beta distributions for leukemia, while parameters y, y, 5, and ¢, if not constant, are assumed to have normal (Gaussian) uncertainty distributions. Parameter (associated regression variable) Cancer type Bur Br y(e) Ha) All solid (0.33 (0.24, 0.47) 0.57 (0.44, 0.74) —0.3 (—0.51, —0.10) —1.4 (—2.2, —0.7) cancers* Thyroid Stomach Colon Leukemia’ 0.53 (0.14, 2.0) 1.05 (0.28, 3.9) 0.21 (0.11, 0.40) 0.48 (0.31, 0.73) 0.63 (0.37, 1.1) —1.1 (0.10, 2.6) 0.43 (0.19, 0.96) —-1.2 (0.10, 2.9) —0.83° —0.3° —0.3° —0.4 (—0.78, 0.0) 0 —1.4° —1.4° 0 (1) 0 ple X t) 0 0 0 0 0 0 0 —0.48 (—1.1, 0.20) 0.42 (0.0, 0.96) “Except thyroid cancer and non-melanomaskin cancer. > Error assumed to be negligible, following BEIR VII (NRC 2006). “ Because dose from fallout was considered to have been received at a low dose rate, the parameter for the dose-squared term in the BEIR VII model for leukemia was set equal to zero.