Effect of radiation exposure on survival after first solid cancer diagnosis in A-bomb survivors

The Life Span Study (LSS) of the Radiation Effects Research Foundation is used to quantify the long-term effects of atomic bomb radiation on cancer incidence and mortality 123 . Recently Brenner et al. [4] updated models of solid cancer incidence and mortality using follow-up through 2009. Since LSS follow-up for mortality begins in 1950, whereas follow-up for incidence begins in 1958, these authors compared cancer mortality using both follow-up from 1950 and follow-up from 1958, the latter done to compare cancer mortality and incidence using identical follow-up periods and identical excess-relative risk (ERR) models. Brenner et al. [4] commented at length about the emerging evidence of an upward curvature in all solid cancer dose response and a sex-specific difference in dose response shape between the incidence and the mortality models. One possible contributing factor to this difference is whether and how pre-diagnosis radiation exposure influences post-cancer diagnosis survival over and above its influence on cancer incidence.

In the LSS analyses of solid cancer incidence and mortality it is not necessarily the case that subjects who die of solid cancers die of the same cancer as originally diagnosed. Whereas solid cancer incidence analysis is based on the first primary solid cancer diagnosis, subsequently diagnosed cancers different from the first primary may be the ultimate cause of death.

One can conjecture about mechanisms by which pre-diagnosis radiation exposure could influence mortality from an incident cancer. Radiation-caused cancer might be genetically and therefore possibly prognostically distinct from spontaneously occurring cancer. Evidence of such genetic differences exists from murine models (e.g., [56] ) and recent human studies [7] . Radiation exposure could alter the immune system [89] , possibly affecting individuals’ innate ability to fight cancer and tolerance of aggressive immunosuppressive treatment for cancer and therefore treatment effectiveness.

Under a null hypothesis that pre-diagnosis radiation exposure does not affect survival after solid cancer diagnosis, one would expect that post-diagnosis survival in individuals who died from the originally diagnosed cancer would not be related to radiation dose. However, post-diagnosis survival of individuals who died of cancers unrelated to the first primary diagnosis could be associated with radiation dose in that radiation exposure is associated with the additional incident cancer [10] which led to the subject’s demise. Post-diagnosis survival in individuals who died of non-cancer-related causes could also be associated with radiation dose through the effect of radiation on non-cancer mortality [1112] .

In this paper we investigate this null hypothesis by analyzing post-diagnosis survival and its relationship to radiation exposure in subjects diagnosed with solid cancers in the LSS from 1958 to 2009 with specific attention to differences in dose response for different causes of death. Bockwoldt et al. [13] recently analyzed post-diagnosis survival in 7728 LSS subjects with gastrointestinal cancer who died of this cause. Their analysis was inconclusive as to the association between pre-diagnosis exposure and post-diagnosis survival. In the current analysis, we include all first primary solid cancers in a single analysis and investigate differences in the association of prior radiation exposure to survival for different causes of death. This approach provides a more sensitive analysis of the association of radiation dose to post-diagnosis survival, especially in those who died from the originally diagnosed cancer.