RSH Data & Documents

"Low Level
Radiation Health
Effects: Compiling
the Data"

Revision 2
March 30, 1999

by Radiation, Science, and
Health, Inc.,
Edited by J. Muckerheide

1.2.1.1
Cancer/Leukemia

"They were working on a sample that was composed of two cohorts of atomic bomb survivors (Hiroshima and Nagasaki) which I name LSS 1950 (Life Span Study, recruited in 1950); they used the dosimetry system named T 65 DR (Tentative Dosimetry published in 1965, revised later). From modified data they published new results with others (Shimizu et al 1987). I name the new sample LSS 1985; the new dosimetry system is named DS 86. I limit this short correspondence to leukemia.

"For LSS 1950, the authors (Kato et al. 1987, p. 647, Table 1) consider only the survivors exposed to less than 0.50 Gy. As I had advised them indirectly (Beebe 1982), they subdivided these survivors into five classes. In spite of that, they can confirm their initial statement (p. 651): 'The frequency of events so far observed among the five dose groups (from 0 to less than 0.50 Gy) varied, but haphazardly, and there were not statistically significant differences among them. Thus, these results failed to suggest the existence of radiation hormesis.'

"Thanks to an earlier publication (Kato and Schull 1980), I enlarge beyond 0.50 Gy the results they discuss. Table I shows an obvious hormesis at Nagasaki (the minimum death rate is 0 in the class beginning at 0.50 Gy); less important at Hiroshima (minimum 17), the hormesis may be easily ignored.

Table I

"Studying LSS 1985 with others (Shimizu et al 1987), they insist:

(1) 'The shape of the dose-response curve is more linear with the new dosimetry in both cities.'

(2) 'Although responses still vary within the low dose range, there is no statistically significant evidence of a positive effect (hormesis) at low doses even when the observations are distributed over finer subdivisions.'

"Such interpretation can discourage researchers and mislead radiation research. I shall comment on them briefly.

(1) Hormesis is effectively reduced in depth in both cities by DS 86 applied on LSS 1985 (Table 1). At Nagasaki it seems narrower; in fact, it is translated towards lower doses (from the 0.20-0.50 Gy interval to 0.05-0.20). Do such changes suffice to justify the use of linear models?

(2) Any hormesis does not appear at Hiroshima with the usual subdivision of the survivors; however, it may be shown with acute leukemia (e.g., see Ichimaru et al 1981, p. 5, Table 2). At Nagasaki it appears in both cases.

(3) Presently, no radiation hormesis has been found deep enough to be statistically significant. Does this justify to put in doubt the hormesis existence?

(4) With the sole publication which gives us the type of each leukemia (Ichimaru et al 1976, Appendix), I have shown hormesis not only at Nagasaki, but also at Hiroshima (Delpha 1978). Later (Delpla 1985), from the same data, I stated that for the acute granulocytic leukemia (Table 2), the decrease of the incidence (hormesis) is statistically 'suggestive' at Hiroshima with p ~6%. At Nagasaki the cohort was not great enough.

…(7) A given hormesis assigns numerical probabilities to two thresholds. With increasing doses, the first one corresponds to the decrease of the natural frequency (protective effect); at the second, the observed frequency exceeds its natural value (pathological effect). I have good hope that researchers will soon confirm statistically (p < 5%) protective effects of adequate doses of ionizing radiation (Delpla 1985; Delpla and Chevalier 1988). "

Table II