March 1993, Vol. 3, No. 1, pp.39-41.

The PSR Quarterly

A Journal of Medicine and Global Survival

Published by Williams and Wilkins for
Physicians for Social Responsibility

Low-Level Radiation

---

Radiation-Induced Cancer from Low-Dose Exposure:   An Independent Analysis, by John W. Gofman, M.D., Ph.D., CNR Book Division, San Francisco, 1990. 480 pp. ($29.95)

---

Large populations all over the globe continue to be exposed to low-level radiation. Sources include natural background radiation, widespread medical uses of ionizing radiation, releases and leakages from nuclear power and weapons manufacturing plants, nuclear tests, reprocessing, and waste storage sites. An added potential hazard to public health in the U.S. stems from government plans to deregulate low-level radioactive waste from uses in industry, research, and hospitals and to allow it to be mixed with general household trash and industrial waste in unprotected dump sites. Long-standing reassurances to the public by official experts have been shown to be false and there is growing demand for critical discussion of the issues. Therefore, Gofman's independent analysis of the health risks of low-dose exposures is especially timely.

        Except for Gofman's new treatise and a few other independent analyses of the mortality statistics among the survivors of the Hiroshima-Nagasaki bombings [1,2], all authoritative radiogenic risk estimates available to the public, to radiation protection experts, and to media reporters originate from tabulations in official reports by official national and international radiation protection commissions who have extensively analyzed the same basic data. The nonspecialist reader of these reports who seeks an explanation of radiation risk estimations and their continuing upward revisions ends up frustrated by the technical jargon and by implicit assumptions built into the mathematical models used in the analysis of primary data. Thus, the reader has no choice but to accept the reports' final conclusions on the experts' authority. For example, most official radiologic risk tabulations for low-dose exposures over long periods of time (the most common mode of exposure of large populations) have been obtained by first calculating those risks from a straight-line-through-zero extrapolation of observed excess cancer mortalities at medium to high doses among the A-bomb survivors. However, before listing these extrapolated values in summary risk tables for low-dose exposures, the numbers have been reduced by postulated "correction factors" (DREPs) from 0.1 to 0.5. Such reductions in radiogenic risk cannot be justified from either recent A-bomb mortality statistics or more recent epidemiologic studies focusing exclusively on populations exposed to low doses, accumulated at low dose-rates [3,4,5]. In his book, Gofman goes to great length to show, consistent with other radiation studies [1-3], that all relevant human data convincingly contradict the hypothesis -- authoritatively stated in much of the current radiation effects literature -- of reduced radiation effects per unit dose at low doses.

        The unique feature of this book is the author's unusual gift of guiding the reader step-by-step through the complexities of evaluating and interpreting statistical data in a nontechnical language, requiring hardly more than a desk calculator and an attentive mind to follow his lucid analysis. Gofman defines all necessary concepts and explicitly discusses the assumptions he makes in reaching his conclusions, leaving the reader free to judge them on their merit. If one accepts Gofman's explicitly stated premises, the evidence presented and the clarity of his logic will convince most readers. The author's major conclusions include:

1. There are independent estimates of radiogenic cancer risks that for medium- to high-dose exposures are consistent with official estimates, but that are substantially higher per unit dose in the low-dose range (below 20 cGy[rad] organ absorbed dose).

2. There is statistically significant evidence for a convex (supralinear, i.e., steeper slope at very low doses) risk versus exposure relation in the low-dose range for the instantaneous A-bomb exposures. A supralinear dose-effect relation is the exact opposite from that implied in official tabulations by applying postulated reduction factors (DREFs) to a no-threshold linear extrapolation from medium- to high-dose risk values. Instead, the data show higher risks at low-dose values than those estimated from a linear extrapolation.

3. A fully documented microdosimetric analysis of the interaction of beta and gamma radiation with tissue cells leads Gofman to the conclusion that the biophysical effects per unit dose that lead to cell mutations are independent of dose rate. In other words, spreading a total dose over longer time periods does not reduce radiogenic risk. This finding is consistent with findings from other epidemiologic studies, limited to low-dose exposures [3,4]. The same discussion includes the biophysical reasons why a widely promoted hypothesis, a threshold dose for radiogenic cancer induction, is untenable.

        Before presenting his radiation risk analysis, Gofman reviews in great detail the recently completed revised dosimetry system (DS86) for the A-bomb survivors, which substantially changed dose assignments to individual survivors. The scientists in charge of the A-bomb study and their scientific advisors (the Hiroshima- based Radiation Effects Research Foundation [RERF]) decided to redefine the cohorts by retaining in the new dosimetry, the arbitrarily chosen former dose ranges. This necessitated a massive shuffling of individual survivors from former into new cohorts and the elimination from the study of 17% of the original subjects (and recorded cancer deaths). One of several consequences of this strategy is that cancer deaths previously recorded for a particular cohort have now been reassigned to another cohort. It is obvious that such shifts or eliminations of cancer mortalities significantly effect resulting risk estimate in a predictable way. Gofman argues strongly that this not only destroys the 41 years continuity of the survivor mortality study, but that it seriously violates accepted principles of "blind" epidemiologic studies, undermining the credibility of RERF's and other official commissions' current statements of radiogenic risks. Gofman demonstrates the efficacy of an alternative methodology that would include the improved dosimetry, yet would preserve continuity of cohorts and comparability with former results. He suggests that mortality follow-up statistics be continued for the original cohorts but to characterize these cohorts by both their (former) mean T65DR doses and their (new) mean DSo6 doses. In short, he recommends to RERF "constant cohort, dual dosimetry" follow-up reports.

        In the closing chapter of the book, "Practical Impacts on Human Health," Gofman compares his own estimate of cancer victims as a result of the Chernobyl disaster with various official estimates. He also provides a unique compilation of quotes on the subject of health effects of low-level radiation from a divergent spectrum of expert opinion, plus an extensive bibliography and a most useful index.

        In his broad discussion of the controversies about methods of analyzing the A-bomb survivor statistics for low-dose radiogenic risk estimations and in his appraisal of that study as the most valuable data base, the author regrettably neglects some recent challenges to the assumption that cancer risks for A-bomb survivors and for a general population are directly comparable [6]. He also implicitly underrates the importance of well-designed long-term mortality studies of workers in nuclear installations for direct radiogenic risk estimates after long-term exposures well below official radiation protection limits [3,4]. These are minor flaws in a unique and most accessible overview of the continuing debate about carcinogenic effects of low levels of ionizing radiation on human health.

-- Rudi H. Nussbaum, Ph.D.