Vol. 324 No. 7
The New England Journal of Medicine
Feb. 14, 1991
HEALTH EFFECTS OF EXPOSURE TO LOW LEVELS OF
IONIZING RADIATION: BEIR V
Prepared by the Committee on the Biological Effects of Ionizing Radiation, National Researdh Council. 421 pp. Washington, D.C., National Academy Press, 1990. $35.
RADIATION-INDUCED CANCER FROM LOW-DOSE
EXPOSURE: AN INDEPENDENT ANALYSIS
By John W. Gofman. 480 pp. San Francisco, Committee for Nuclear Responsibility Book Division, 1990. $29.95.
Two national advisory groups have great influence with regard to the safe conduct of the population through an environment contaminated with ionizing radiation. These are the National Council on Radiation Protection and Measurements and the National Research Council's Committee on the Biological Effects of ionizing Radiation (BEIR). Over the years, both these groups have raised their estimates of the risk of radiation-induced cancer as new evidence has accumulated on the delayed adverse effects of low-level exposure. Now comes a book published by an independent education group (the Committee for Nudear Responsibility) that takes strong issue with the most recent report of BEIR (BEIR V). The author, John W. Gofman, is the founder and former director of the Biomedical Research Division of the Lawrence Livermore National Laboratory.
Both these works agree that previous assessments of the dangers of radiation underestimated the risk, but they reach substantialy different conclusions about the magnitude of the risk, especially when the radiation is at lower doses (below 10 rem) and the doses are delivered slowly. Both reports primarily concern ionizing radiations with a low linear energy transfer, such as gamma rays or x-rays, as opposed to radiations with a high linear energy transfer, such as neutrons or alpha particles. We compare some of the features and major conclusions of these books.
Beginning in 1950, more than 90,000 atomic-bomb survivor from Hiroshima and Nagasaki were enrolled in a lifetime health study. The Radiation Effects Research Foundation--an agency sponsored jointly by the U.S. and Japanese governments--has been in charge of this study since 1975. Its data provide direct quantitative evidence of radiation-induced cancer from short-term exposure of organs at doses of 11 to 15 rem. This prospective study is the cornerstone of the epidemiologic evidence concerning the effects of radiation on humans. A substantial body of information about the health and mortality of the atomic-bomb survivors is in hand. Most of the people exposed at an early age are still living; their ultimate fate will provide critical new data in the ongoing analysis. Recently, the Radiation Effects Research Foundation altered the architecture of this study in major ways to account for new dose estimations, shifting thousands of survivors into different cohort groups and temporarily dropping about 15,000 survivors from the study because of "dose uncertainties."
During the past 40 years, various research organizations, committea, and governmental agencies have evaluated the atomic-bomb study and others in humans, plus data in animals, in assessing the consequences and deriving estimates of risk from exposure to ionizing radiation. Cancers, leukemias, and genetic effects have all been demonstrated to result from both short-term and long-term exposure. Over time, the growing body of scientific evidence showing that radiation is more hazardous than previously thought has resulted in upward revisions of the estimates of the risk of cancer. Since the guidelines for allowable or permissible levels of exposure to low-level ionizing radiation are based on these risk estimates, their accuracy has major public health implications.
The BEIR V document evaluates several aspects of the effects of low-level radiation on humans and animals, including the induction of leukemia, the induction of cancer both generally and at specific sites, genetic effects, and the effects of in utero exposure on brain development and childhood cancers. Three large chapters examine the induction of cancer and leukemia and formulate assessments of risk. The other chapters cover scientific principles and background information, genetic effects, other somatic and fetal effects, epidemiologic studies involving low doses of radiation, and data and analysis pertaining to research in animals. The text is well written, well organized, and extensively referenced. The executive summary outlines the major conclusions clearly. Certain sections on mathematical risk models are complex. Unfortunately, the glossary and index are incomplete, weakening the overall presentation and the reader's ability to find information quickly. For example, "dose-rate effectiveness factor" is an important concept in this work, and although we found it mentioned or discussed at least 17 times in the text, the index only noted 2 of the minor mentions. Some other key words and concepts are not indexed at all.
The second of these books, that by Gofman, focuses almost exclusively on the induction of cancer in humans as a result of low-level ionizing radiation. The book is well organized, clear, exhaustively detailed, and comprehensively referenced. As a result, lay persons or students of other disciplines will be able to master the information with some effort. Through the use of raw data, graphs, tables, charts, and calculations, the reader is taken step by step through the complexities of physics, statistics, and epidemiology. Some sections are highly technical. The book is organized into 25 major chapters, each of which lays the scientific foundation for the next, into which it flows, although some chapters could also stand alone. The 12 supporting chapters provide additional analysis or examples of key points made in the main body of the book. There are frequent cross-references from one section to another. Extensive direct quotations from other reports facilitate an understanding of the views of other analysts. The "index and glossary" is one of the most comprehensive and thoughtful we have seen--brief definitions often appear with the index entry, flagged entries locate the meaning of a term or phrase in context, and even images and phrases have their own entria.
Some of Gotman's major conclusions about the induction of cancer from low-level ionizing radiation are that (1) there are adequate human epidemiologic data on the effects of radiation at low doses to quantify risks directly at those dose levels, without extrapolating from studies of high doses; (2) there is no safe dose or dose-rate -- i.e., there is no threshold below which there is no risk; (3) there is no protection offered from fractionation or the slow delivery of low total doses -- i.e., dose-rate-effectiveness factors, which predict decreased risk under these slow-dose circumstances, should not be used for humans; (4) in the low- dose range, the risk of cancer is possibly more severe per dose-unit than in the moderate- and high-dose ranges -- i.e., the dose-response curve may be supralinear; (5) the approximate lifetime yield of fatal cancer in the low-dose range is 27 excess deaths from cancer per 10,000 person-rem (wholebody dose) in populations of mixed ages, but for young persons the risk is even higher; (6) over the course of several decades, about 400,000 people in Europe and the Soviet Union combined could die of cancer resulting from long-term exposure to fallout from Chernobyl, and (7) there is no scientific validation for the concept of hormesis (a net beneficial effect from radiation).
Gofman devotes 13 chapters to a detailed analysis of the atomic-bomb data base, and he relies heavily on those findings and other evidence in humans in deriving the conclusions listed above. As part of this process, he presents the raw data on mortality that were accumulated from 1950 to 1982 for the survivors of Hiroshima and Nagasaki. Although he is sharply critical of the ways in which the Radiation Effects Research Foundation is retroactively altering the atomic-bomb study (e.g., dismantling cohort groups and creating new ones) in order to account for new dose estimates, Gofman supports the use of improved dosimetry. He demonstrates the effect of a simple method of parallel analysis that he calls "constant-cohort, dual-dosimetry," which allows the incorporation of new dose estimates but leaves the original prospective architecture and cohort groups of the study intact. He pleads that failure to preserve continuity in this "uniguely valuable database" will invalidate its legitimacy as a true prospective epidemiologic study and throw into question the reliability of future results.
By contrast, some of the major conclusions of BEIR V about the effects of low-level ionizing radiation are that (1) there are insufficient epidemiologic data at low doses to quantify directly the risk of cancer in humans at those levels, and extrapolation from higher doses (above 10 rem) is necessary; (2) epidemiologic data cannot exclude the existence of a threshold in the millisievert dose range (1 millisievert equals 0.1 rem), and therefore the possibility cannot be ruled out that there are no risks from exposures comparable to the natural background level; (3) for low doses of radiation with a low linear energy transfer delivered slowly, the lifetime risk is less, "possibly by a dose-rate-effectiveness factor of 2 or more"; (4) for cancer other than leukemia, the dose-response curve is linear throughout the dose range under 400 rem, and for leukemia it is linear quadratic; (5) the approximate lifetime yield of fatal cancer (assuming short-term 10-rem whole body exposure to gamma rays per person) is eight excess deaths from cancer per 10,000 person-rem for populations of mixed ages, but for children the risk is probably twice as high; (6) in utero exposure can cause childhood cancers and leukemias, and possibly disease in adulthood; and (7) the most sensitive gestational age for radiation-induced mental retardation is 8 to 15 weeks, with the risk being a 4 percent chance of retardation per 10 rem of exposure.
Although the findings (and methods) of these two reports differ on major points, there are substantial areas of agreement. Both find radiation more hazardous than was previously believed. Both find that the dose-dependent excess of cancer is best expressed with a "relative" risk estimate or model (i.e., "the number of excess cancers per unit dose induced by radiation is increased with attained age, while the risk of radiogenic cancer relative to the spontaneous incidence remains comparatively constant"). Both find that there is necessarily some uncertainty and imprecision in their risk estimates. They agree that with the completion (in a few decades) of the atomic-bomb study, a more precise estimate of the survivors' lifetime risk will emerge, and that future modifications of the risk will be made as more data from all sources become available. They find children at higher risk per dose-unit of radiation. Both indicate that x-rays (from medical exposures or other sources of x-rays) may be twice as potent a carcinogen as the comparable dose of gamma rays and that therefore their risk values may need to be doubled when the effects of x-rays are predicted. Neither finds scientific evidence to support the hypothesis of hormesis.
One might ask why continuing evaluations of the effects of low-level ionizing radiation are important. To take only one example, a former chairman of the International Commission on Radiological Protection indicated some 12 years ago that if the permissible occupational exposure were to be reduced by a factor of 10 (i.e., from 5 to 0.5 rem per year), he doubted whether the nuclear-power plants of the time would have been able to continue operations. The implications of making regulations that meet scientific and health standards become obvious.
We would like to examine the forecast of fatal cancer derived from both these reports, when it is applied to industry standards for protection from radiation in the past and the present. With either analysis, it appears that even the current permissible exposure of 5 rem of whole-body radiation per year for nuclear-power workers is not actually a "safe" dose. What, then, does a permissible dose of radiation really mean? Warren Sinclair, president of the National Council on Radiation Protection and Measurements, recently said that the current permissible limits "were likely to be reduced" because of the new BEIR report.
First, consider that in 1934 the International Commission on Radiological Protection proposed a 52-roentgen (1 roentgen equals about 0.88 rem, therefore 52 roentgens equal 46 rem) maximal permissible yearly whole-body radiation exposure for workers -- a standard the experts believed was safe. This standard was "used world-wide until 1950." With the BEIR V data, one arrives at a prediction of one extra death from cancer per 3588 person-rem exposure to low-level ionizing radiation (after the application of a dose-rate-effectiveness factor of 2 and adjustment of the risk values for a population of workers 18 to 65 years of age). Therefore, in a population of 3588 radiation workers who received this maximal permissible dose in one year, 46 extra fatal cancers might occur. The same per annum exposure for 16 years (1934 to 1950) could eventually result in the occurrence of 736 extra cancers in the same population. With Gofman's estimate of a cancer risk that is 3.83 times higher than the BEIR V estimate (with correction for dose-rate-effectiveness factor), 2819 workers of an original group of 3588 would have received doses of radiation causing fatal cancer in the 16-year period, if they had been exposed to the maximal amount permissible every year. The spread of potential fatality rates is certainly impressive.
Second, today's worker in an environment where radiation is present is allowed a maximum of 5 rem of whole-body exposure per annum. If 3588 workers received this dose slowly in one year, the BEIR V data would allow a prediction of 5 future excess deaths from cancer, whereas the Gofman method would predict 19.
Gofman and the BEIR V committee have each produced a fascinating document. They analyzed many of the same data but arrived at different conclusions. Although BEIR V finds acute exposure to low-level ionizing radiation to be about three times more hazardous as a cause of excess deaths from cancer than was estimated by the BEIR III committee a decade ago, Gofman condudes that the new BEIR V calculations still underestimate the risk substantially.
We strongly recommend both these excellent and timely books for physicians, engineers, and public health officials concerned with radiation, the environment, and public health. As humans contemplate prolonged flight beyond the magnetosphere, in the intense radiation environment of the nearby solar system, a whole new generation of space-flight engineers, physicians, and safety offficers must become deeply involved in this process.
G. THEODORE DAVIS, M.D.
ANDRE J. BRUWER, M.D.
1010 Las Lomas N.E.
Albuquerque, NM 87102