206
Health Physics
ERRfor protracted exposure 1s assumed to be proportional to radiation dose m Gy, and the ERR per Gy 1s
expressed as
ERR Gy"! =B8X explyXe+8Xt+pXeXd
4
Here, 8 and ¢ are additional uncertain parameters, and
t = In[(attained age — exposure age) X 257']
(3)
The parameters im equations (1) and (4) are
assumed to be random vanables 8, and Bp are
distributed as lognormal for solid cancers and beta for
August 2010, Volume 99, Number 2
mean = GM X exp[0 5 x In°(GSD)]
= 4678 X 1 2533 =5 862
For an uncertain dose estrmate of 0 04 Gy (0 01 Gy from
external radiation and 0 03 Gy from mternal radiation as
discussed above under “Radiation dose”), the estimated
ERR mustreflect the statistically mdependent uncertamtes of both the estrmated ERR per Gy and the estrmated
dose m Gy Given a lognormal uncertamty distribution
for the estrmated dose with GM = 002558 Gy and
GSD = 2 51 (see “Radiation dose” above), the estimated
ERR at mean dose 0 04 Gy, then, 1s considered to be
approximately lognormal with
GM = 002558 Gy X 4678 Gy"! =0 1197,
leukemia, and y, 4, 5, and g are normal or constant
Parameter medians and 95% uncertainty lymits are
presented in Table 3 for all solid cancers less thyroid
(9)
and
BEIR VII dose-response estimates, 1t differs from BEIR
mean = GM X exp[0 5 X In{GSD)]
VII with respect to modification of dose response at low
doses and low dose rates (as discussed below under
“Adjustment for protracted exposures”), transfer between
the Japanese A-bomb survivors and the exposed Marshall Islands population (as discussed above under “Estimation of baseline cancer rates” and below under
“Transfer of estrmated excess risk to the exposed MI
populations”), and to treatment of the latent period
between radiation exposure and diagnosis of cancer (as
discussed below under “Latent period”) In these matters,
we followed an earlier National Institutes of Health
approach (NIH 1985, 2003) to evaluate the extent to
which a given cancer diagnosis might beattributable to a
given prior history of exposure to 1omzing radiation
Example. From Table 3, the estimated ERR per Gy,
accordmg to BEIR VII, for radiation-related thyroid
cancer ma woman at age 62 y, following exposure at age
(8)
GSD = exp{[In°(2 51) + n° 96)]7} = 3 127,
and non-melanoma skin cancer, for leukemia, and for
cancers of the thyroid gland, stomach, and colon
While the approach used mthe present analysis uses
(7)
= 01197 X exp[0 5 X In°3. 127)] = 0229
(10)
Adjustment for protracted exposures
In the BEIR VII report (NRC 2006), as elsewhere, a
Imear-quadratic dose-response model 1s used for leuke-
mia ERR associated with an acute radiation dose, but for
protracted doses, the coefficient for dose-squared 1s set
equal to zero, givmg a lmear dose-response model for
leukerma For solid cancer msk followmg protracted or
very low-dose exposures, the ERR 1s divided by a
dose-and-dose-rate effectiveness factor (DDREF) The
present calculations mvolve a different DDREF, shown
m Fig 4 (left panel), which was developed for the
“Interactive Radio-Epidemiological Program” (IREP)
used to facilitate adjudication of compensation claims
agamst the US government for radiation-related cancers
(NIH 2003, Kocher et al 2008) When applied to the
thyroid cancer example mtroduced under “Models for
estrmation of radiation-related cancer nsk,” the uncer-
12 y, 1s
tamty distribution for the ERR estimate divided by the
ERR Gy7! = 1 05 X exp[—0 83 x (12 — 30) x 107!
sponds closely to a lognormal distribution with GM =
—00X In(62 x 60°'}] = 4678
DDREF, evaluated by Monte Carlo simulation, corre-
(6)
(Note that the ERR for excess thyroid cancer, unlike that
for other cancers, does not depend upon attamed age )
Wetreat this esate as an uncertam value distributed as
approximately lognormal with geometric mean GM =
468 and GSD = 96(as indicated by the 95% uncertamty bounds for the parameter 8, m Table 3) The
arithmetic mean of this distribution 1s
0 0805 and GSD = 3 40 (mean = 0170) (Fig 5)
Latent period
As shown m Table 3, the dose-specific ERR may
depend upon attaimed age and/or time following exposure, but experimental studies at the cellular and animal
level strongly suggest that the process of radiation
carcmogenesis requires me, 1¢, there 1s a mmmum
latent period of uncertam duration that 1s superrmposed