Cancer incidence and mortality after low-dosage radiation exposure: Epidemiological aspects

Current recommendations for limiting exposure to ionizing radiation are based on the linear no-threshold (LNT) model for radiation carcinogenesis under which every dose, no matter how low, bears some cancer risk. In this review, epidemiological evidence is discussed that the LNT hypothesis is incorrect at low doses. A large set of data was accumulated that show that cancer risk after ordinarily encountered radiation exposure (natural background radiation, medical X-rays, etc.) is much lower than estimates based on the LNT model. The discovery of low-level radiation hormesis (stimulating effect) implies a non-linear dose-response curve in the low-dosage region. Further studies in this field will provide new insights into the mechanisms of radiation carcinogenesis.     

Roles of stem cells in tissue turnover and radiation carcinogenesis

Radiation research has its foundation on the target and hit theories, which assume that the initial stochastic deposition of energy on a sensitive target in a cell determines the final biological outcome. This assumption is rather static in nature but forms the foundation of the linear no-threshold (LNT) model of radiation carcinogenesis. The stochastic treatment of radiation carcinogenesis by the LNT model enables easy calculation of radiation risk, and this has made the LNT model an indispensable tool for radiation protection. However, the LNT model sometimes fails to explain some of the biological and epidemiological data, and this suggests the need for insight into the mechanisms of radiation carcinogenesis. Recent studies have identified unique characteristics of the tissue stem cells and their roles in tissue turnover. In the present report, some important issues of radiation protection such as the risk of low-dose-rate exposures and in utero exposures are discussed in light of the recent advances of stem cell biology.     

Low-dose ionizing radiation: induction of differential intracellular signalling possibly affecting intercellular communication

Given the complexity of the carcinogenic process and the lack of any mechanistic understanding of how ionizing radiation at low-level exposures affects the multistage, multimechanism processes of carcinogenesis, it is imperative that concepts and paradigms be reexamined when extrapolating from high dose to low dose. Any health effect directly linked to low-dose radiation exposure must have molecular/biochemical and biological bases. On the other hand, demonstrating some molecular/biochemical or cellular effect, using surrogate systems for the human being, does not necessarily imply a corresponding health effect. Given the general acceptance of an extrapolated LNT model, our current understanding of carcinogenesis cries out for a resolution of a real problem. How can a low-level acute, or even a chronic, exposure of ionizing radiation bring about all the different mechanisms (mutagenic, cytotoxic, and epigenetic) and genotypic/phenotypic changes needed to convert normal cells to an invasive, malignant cell, given all the protective, repair, and suppressive systems known to exist in the human body? Until recently, the prevailing paradigm that ionizing radiation brings about cancer primarily by DNA damage and its conversion to gene and chromosomal mutations, drove our interpretation of radiation carcinogenesis. Today, our knowledge includes the facts both that epigenetic events play a major role in carcinogenesis and that low-dose radiation can also induce epigenetic events in and between cells in tissues. This challenges any simple extrapolation of the LNT model. Although a recent delineation of “hallmarks” of the cancer process has helped to focus on how ionizing radiation might contribute to the induction of cancers, several other hallmarks, previously ignored—namely, the stem cells in tissues as targets for carcinogenesis and the role of cell–cell communication processes in modulating the radiation effects on the target cell—must be considered, particularly for the adaptive response, bystander effects, and genomic instability phenomena.     

Cancer-prone mice expressing the Ki-rasG12C gene show increased lung carcinogenesis after CT screening exposures

A >20-fold increase in X-ray computed tomography (CT) use during the last 30 years has caused considerable concern because of the potential carcinogenic risk from these CT exposures. Estimating the carcinogenic risk from high-energy, single high-dose exposures obtained from atomic bomb survivors and extrapolating these data to multiple low-energy, low-dose CT exposures using the Linear No-Threshold (LNT) model may not give an accurate assessment of actual cancer risk. Recently, the National Lung Cancer Screening Trial (NLST) reported that annual CT scans of current and former heavy smokers reduced lung cancer mortality by 20%, highlighting the need to better define the carcinogenic risk associated with these annual CT screening exposures. In this study, we used the bitransgenic CCSP-rtTA/Ki-ras mouse model that conditionally expresses the human mutant Ki-ras(G12C) gene in a doxycycline-inducible and lung-specific manner to measure the carcinogenic risk of exposure to multiple whole-body CT doses that approximate the annual NLST screening protocol. Irradiated mice expressing the Ki-ras(G12C) gene in their lungs had a significant (P = 0.01) 43% increase in the number of tumors/mouse (24.1 ± 1.9) compared to unirradiated mice (16.8 ± 1.3). Irradiated females had significantly (P < 0.005) more excess tumors than irradiated males. No tumor size difference or dose response was observed over the total dose range of 80-160 mGy for either sex. Irradiated bitransgenic mice that did not express the Ki-ras(G12C) gene had a low tumor incidence (≤ 0.1/mouse) that was not affected by exposure to CT radiation. These results suggest that (i) estimating the carcinogenic risk of multiple CT exposures from high-dose carcinogenesis data using the LNT model may be inappropriate for current and former smokers and (ii) any increased carcinogenic risk after exposure to fractionated low-dose CT-radiation may be restricted to only those individuals expressing cancer susceptibility genes in their tissues at the time of exposure.     

Molecular biology, epidemiology, and the demise of the linear no-threshold (LNT) hypothesis

The prime concern of radiation protection policy since 1959 has been protecting DNA from damage. The 1995 NCRP Report 121 on collective dose states that since no human data provides direct support for the linear no threshold hypothesis (LNT), and some studies provide quantitative data that, with statistical significance, contradict LNT, ultimately, confidence in LNT is based on the biophysical concept that the passage of a single charged particle could cause damage to DNA that would result in cancer. Current understanding of the basic molecular biologic mechanisms involved and recent data are examined before presenting several statistically significant epidemiologic studies that contradict the LNT hypothesis. Over eons of time a complex biosystem evolved to control the DNA alterations (oxidative adducts) produced by about 10(10) free radicals/cell/d derived from 2-3% of all metabolized oxygen. Antioxidant prevention, enzymatic repair of DNA damage, and removal of persistent DNA alterations by apoptosis, differentiation, necrosis, and the immune system, sequentially reduce DNA damage from about 10(6) DNA alterations/cell/d to about 1 mutation/cell/d. These mutations accumulate in stem cells during a lifetime with progressive DNA damage-control impairment associated with aging and malignant growth. A comparatively negligible number of mutations, an average of about 10(-7) mutations/cell/d, is produced by low LET radiation background of 0.1 cGy/y. The remarkable efficiency of this biosystem is increased by the adaptive responses to low-dose ionizing radiation. Each of the sequential functions that prevent, repair, and remove DNA damage are adaptively stimulated by low-dose ionizing radiation in contrast to their impairment by high-dose radiation. The biologic effect of radiation is not determined by the number of mutations it creates, but by its effect on the biosystem that controls the relentless enormous burden of oxidative DNA damage. At low doses, radiation stimulates this biosystem with consequent significant decrease of metabolic mutations. Low-dose stimulation of the immune system may not only prevent cancer by increasing removal of premalignant or malignant cells with persistent DNA damage, but used in human radioimmunotherapy may also completely remove malignant tumors with metastases. The reduction of gene mutations in response to low-dose radiation provides a biological explanation of the statistically significant observations of mortality and cancer mortality risk decrements, and contradicts the biophysical concept of the basic mechanisms upon which, ultimately, the NCRPs confidence in the LNT hypothesis is based.     

Predicting cancer rates in astronauts from animal carcinogenesis studies and cellular markers

The radiation space environment includes particles such as protons and multiple species of heavy ions, with much of the exposure to these radiations occurring at extremely low average dose-rates. Limitations in databases needed to predict cancer hazards in human beings from such radiations are significant and currently do not provide confidence that such predictions are acceptably precise or accurate. In this article, we outline the need for animal carcinogenesis data based on a more sophisticated understanding of the dose-response relationship for induction of cancer and correlative cellular endpoints by representative space radiations. We stress the need for a model that can interrelate human and animal carcinogenesis data with cellular mechanisms. Using a broad model for dose-response patterns which we term the "subalpha-alpha-omega (SAO) model", we explore examples in the literature for radiation-induced cancer and for radiation-induced cellular events to illustrate the need for data that define the dose-response patterns more precisely over specific dose ranges, with special attention to low dose, low dose-rate exposure. We present data for multiple endpoints in cells, which vary in their radiosensitivity, that also support the proposed model. We have measured induction of complex chromosome aberrations in multiple cell types by two space radiations, Fe-ions and protons, and compared these to photons delivered at high dose-rate or low dose-rate. Our data demonstrate that at least three factors modulate the relative efficacy of Fe-ions compared to photons: (i) intrinsic radiosensitivity of irradiated cells; (ii) dose-rate; and (iii) another unspecified effect perhaps related to reparability of DNA lesions. These factors can produce respectively up to at least 7-, 6- and 3-fold variability. These data demonstrate the need to understand better the role of intrinsic radiosensitivity and dose-rate effects in mammalian cell response to ionizing radiation. Such understanding is critical in extrapolating databases between cellular response, animal carcinogenesis and human carcinogenesis, and we suggest that the SAO model is a useful tool for such extrapolation.     

Radon sources and impacts: a review of mining and non-mining issues

Radon is a ubiquitous natural carcinogen derived from the three primordial radionuclides of the uranium series (²³⁸U and ²³⁵U) and thorium series (²³²Th). In general, it is present at very low concentrations in the outdoor or indoor environment, but a number of scenarios can give rise to significant radiological exposures. Historically, these scenarios were not recognised, and took many centuries to understand the links between the complex behaviour of radon and progeny decay and health risks such as lung cancer. However, in concert with the rapid evolution in the related sciences of nuclear physics and radiological health in the first half of the twentieth century, a more comprehensive understanding of the links between radon, its progeny and health impacts such as lung cancer has evolved. It is clear from uranium miner studies that acute occupational exposures lead to significant increases in cancer risk, but chronic or sub-chronic exposures, such as indoor residential settings, while suggestive of health risks, still entails various uncertainties. At present, prominent groups such as the BEIR or UNSCEAR committees argue that the ‘linear no threshold’ (LNT) model is the most appropriate model for radiation exposure management, based on their detailed review and analysis of uranium miner, residential, cellular or molecular studies. The LNT model implies that any additional or excess exposure to radon and progeny increases overall risks such as lung cancer. A variety of engineering approaches are available to address radon exposure problems. Where high radon scenarios are encountered, such as uranium mining, the most cost effective approach is well-engineered ventilation systems. For residential radon problems, various options can be assessed, including building design and passive or active ventilation systems. This paper presents a very broad but thorough review of radon sources, its behaviour (especially the importance of its radioactive decay progeny), common mining and non-mining scenarios which can give rise to significant radon and progeny exposures, followed by a review of associated health impacts, culminating in typical engineering approaches to reduce exposures and rehabilitate wastes.     

Estimating radiation-induced cancer risks at very low doses: rationale for using a linear no-threshold approach

The possible cancer risks caused by ionizing radiation doses of ~1 mSv or less are too small to be estimated directly from epidemiological data. The linear no-threshold (LNT) approach to estimating such risks involves using epidemiological data at higher (but still low) doses to establish an “anchor point”, and then extrapolating the excess cancer risk linearly down from this point to the low dose of interest. The study in this issue by Professor Tubiana and colleagues, summarizing a French Academy of Sciences report, argues that such LNT extrapolations systematically give substantial overestimates of the excess cancer risk at very low doses. We suggest that, to the contrary, even if there are significant deviations from linearity in the relevant dose range, potentially caused by the effects of inter-cellular interactions or immune surveillance, we know almost nothing quantitatively about these effects. Consequently, we do not know the magnitude, nor even the direction of any such deviations from linearity—the risks could indeed be lower than those predicted by a linear extrapolation, but they could well be higher.     

Linear-no-threshold is a radiation-protection standard rather than a mechanistic effect model

The linear-no-threshold (LNT) controversy covers much more than the mere discussion whether or not “the LNT hypothesis is valid”. It is shown that one cannot expect to find only one or even the only one dose–effect relationship. Each element within the biological reaction chain that is affected by ionizing radiation contributes in a specific way to the final biological endpoint of interest. The resulting dose–response relationship represents the superposition of all these effects. Till now there is neither a closed and clear picture of the entirety of radiation action for doses below some 10 mSv, nor does clear epidemiological evidence exist for an increase of risk for stochastic effects, in this dose range. On the other hand, radiation protection demands for quantitative risk estimates as well as for practicable dose concepts. In this respect, the LNT concept is preferred against any alternative concept. However, the LNT concept does not necessarily mean that the mechanism of cancer induction is intrinsically linear. It could hold even if the underlying multi-step mechanisms act in a non-linear way. In this case it would express a certain “attenuation” of non-linearities. Favouring LNT against threshold-, hyper-, or sub-linear models for radiation-protection purposes on the one hand, but preferring one of these models (e.g. for a specific effect) because of biological considerations for scientific purposes on the other hand, does not mean a contradiction.     

Enhanced intestinal tumor multiplicity and grade in vivo after HZE exposure: mouse models for space radiation risk estimates

Carcinogenesis induced by space radiation is considered a major risk factor in manned interplanetary and other extended missions. The models presently used to estimate the risk for cancer induction following deep space radiation exposure are based on data from A-bomb survivor cohorts and do not account for important biological differences existing between high-linear energy transfer (LET) and low-LET-induced DNA damage. High-energy and charge (HZE) radiation, the main component of galactic cosmic rays (GCR), causes highly complex DNA damage compared to low-LET radiation, which may lead to increased frequency of chromosomal rearrangements, and contribute to carcinogenic risk in astronauts. Gastrointestinal (GI) tumors are frequent in the United States, and colorectal cancer (CRC) is the third most common cancer accounting for 10% of all cancer deaths. On the basis of the aforementioned epidemiological observations and the frequency of spontaneous precancerous GI lesions in the general population, even a modest increase in incidence by space radiation exposure could have a significant effect on health risk estimates for future manned space flights. Ground-based research is necessary to reduce the uncertainties associated with projected cancer risk estimates and to gain insights into molecular mechanisms involved in space-induced carcinogenesis. We investigated in vivo differential effects of γ-rays and HZE ions on intestinal tumorigenesis using two different murine models, ApcMin/+ and Apc1638N/+. We showed that γ- and/or HZE exposure significantly enhances development and progression of intestinal tumors in a mutant-line-specific manner, and identified suitable models for in vivo studies of space radiation–induced intestinal tumorigenesis.     

Radiation-induced carcinogenesis: mechanistically based differences between gamma-rays and neutrons, and interactions with DMBA

Different types of ionizing radiation produce different dependences of cancer risk on radiation dose/dose rate. Sparsely ionizing radiation (e.g. γ-rays) generally produces linear or upwardly curving dose responses at low doses, and the risk decreases when the dose rate is reduced (direct dose rate effect). Densely ionizing radiation (e.g. neutrons) often produces downwardly curving dose responses, where the risk initially grows with dose, but eventually stabilizes or decreases. When the dose rate is reduced, the risk increases (inverse dose rate effect). These qualitative differences suggest qualitative differences in carcinogenesis mechanisms. We hypothesize that the dominant mechanism for induction of many solid cancers by sparsely ionizing radiation is initiation of stem cells to a pre-malignant state, but for densely ionizing radiation the dominant mechanism is radiation-bystander-effect mediated promotion of already pre-malignant cell clone growth. Here we present a mathematical model based on these assumptions and test it using data on the incidence of dysplastic growths and tumors in the mammary glands of mice exposed to high or low dose rates of γ-rays and neutrons, either with or without pre-treatment with the chemical carcinogen 7,12-dimethylbenz-alpha-anthracene (DMBA). The model provides a mechanistic and quantitative explanation which is consistent with the data and may provide useful insight into human carcinogenesis.     

Analysis of thyroid cancer data from the Ukraine after ’Chernobyl’ using a two-mutation carcinogenesis model

The thyroid cancer data of children in the northern regions of the Ukraine after the reactor accident at Chernobyl were combined with thyroid dose measurements in the same regions and analysed using a two- mutation carcinogenesis model. The best fit was obtained for radiation acting as an initiating agent, i.e. on the first mutation of the model. The observed relatively high increase of thyroid cancer incidence after 1990 in children exposed to radiation released after the reactor accident could be ascribed to the high thyroid doses and the relatively low background thyroid cancer incidence in children. The maximum annual incidence is predicted to occur fairly soon after the reactor accident, i.e. about 10 years. For adults, the predicted relative increase of annual thyroid cancers is much lower than for children younger than 20 years. The modelling results are used to derive risk estimates for radiation-induced thyroid cancer. These risk estimates are dependent on age at exposure, follow-up time and the background thyroid cancer incidence. The calculated excess absolute risk for a population of all ages is about one-third of that currently used by ICRP, but for children the calculated absolute risks are about a factor of 3 higher than derived in other epidemiological studies. The model results indicate that the excess absolute radiation risk per unit dose for children is about the same as or a little lower than that for adults. Received: 11 May 1999 / Accepted: 30 December 1999     

Recent reports on the effect of low doses of ionizing radiation and its dose–effect relationship

Recently, the risk associated with low doses of ionizing radiation has gained new interest. Here, we analyze and discuss the major differences between two reports recently published on this issue; the report of the French Academy of Sciences and of the French Academy of Medicine published in March 2005, and the BEIR VII—Phase 2 Report of the American National Academy of Sciences published as a preliminary version in July 2005. The conclusion of the French Report is that the linear no-threshold relationship (LNT) may greatly overestimate the carcinogenic effect of low doses (<100 mSv) and even more that of very low doses (<10 mSv), such as those delivered during X-ray examinations. Conversely, the conclusion of the BEIR VII report is that LNT should be used for assessing the detrimental effects of these low and very low doses. The causes of these diverging conclusions should be carefully examined. They seem to be mostly associated with the interpretation of recent biological data. The point of view of the French Report is that these recent data are incompatible with the postulate on which LNT is implicitly based, namely the constancy of the carcinogenic effect per unit dose, irrespective of dose and dose rate.     

Lack of nontargeted effects in murine bone marrow after low-dose in vivo X irradiation

Exposure to high doses of ionizing radiation unequivocally produces adverse health effects including malignancy. At low doses the situation is much less clear, because effects are generally too small to be estimated directly by epidemiology, and extrapolation of risk and establishment of international rules and standards rely on the linear no-threshold (LNT) concept. Claims that low doses are more damaging than would be expected from LNT have been made on the basis of in vitro studies of nontargeted bystander effects and genomic instability, but relevant investigations of primary cells and tissues are limited. Here we show that after low-dose low-LET in vivo radiation exposures in the 0-100-mGy range of murine bone marrow there is no evidence of a bystander effect, assessed by p53 pathway signaling, nor is there any evidence for longer-term chromosomal instability in the bone marrow at doses below 1000 mGy. The data are not consistent with speculations based on in vitro nontargeted effects that low-dose X radiation is more damaging than would be expected from linear extrapolation.     

Multistage models and the incidence of cancer in the cohort of atomic bomb survivors

The analyses in this paper show that a number of biologically based models describe cancer incidence among the A-bomb survivors equally well. However, these different models can predict very different temporal patterns of risk after irradiation. No evidence was found to support the previous claim of Pierce and Mendelsohn that excess cancer risks for the solid tumors depend only upon attained age and not on age at exposure or time since exposure. Although the A-bomb survivor cohort is the largest epidemiological data set for the study of radiation and cancer, it is not large enough to discriminate among various possible carcinogenic mechanisms. Unfortunately for hypothesis generation, the data appear to be consistent with a number of different mechanistic interpretations of the role of radiation in carcinogenesis.     

Is there a real risk of radiation-induced breast cancer for postmenopausal women?

The risk of radiation-induced breast cancer decreases with increasing age at exposure. Thus, for calculating the individual risk for a patient undergoing mammography, age-related risk coefficients need to be used. In this report, the results of epidemiological studies on risks of radiation-induced breast cancer are reviewed indicating that the available data do not show the risk to be enhanced for women exposed at the age of 55 years or older. This lack of evidence is reflected by the fact that the risk coefficients recommended by national and international advisory bodies differ by a factor of 10 or more for age at exposure of 50–60 years or older. A hypothesis is proposed indicating that the risk of radiation-induced breast cancer might decrease considerably at the time of menopause. The hypothesis is based on the following line of arguments: (1) evidence has accumulated from molecular genetic studies indicating that the development of colorectal cancer requires a cascade of subsequent mutations consisting of at least seven genetic events. (2) For colorectal cancer, the annual rates of incidence and mortality increase with age to the power of 5–6. Thus, the number of mutation steps (minus 1) is approximately reflected by the power of age dependence. (3) For western populations, the incidence and mortality of breast cancer up to the age of about 50 years increase with age to the power of about 6, indicating that a similar number of genetic events might be involved in development of breast cancer as has been identified for colorectal cancer. (4) For women aged 50 years or older, breast cancer occurs at an annual rate that is about proportional to age or age squared. This may mean that after menopause, the processes in the multistep mutation cascade leading to breast cancer are slowed down by a factor of about 4 or more. (5) The constant relative risk model of radiation carcinogenesis implies for solid cancers that radiation acts by inducing additional mutations in the earlier steps of the multistep cascade. It is suggested that the break-point in the age-specific annual rate of breast cancer incidence at menopause is associated with a corresponding drop in radiation sensitivity with respect to induction of breast cancer. Received: 8 January 2001 / Accepted: 20 March 2001     

Calculation of the 1995 lung cancer incidence in the Netherlands and Sweden caused by smoking and radon: risk implications for radon

A two-mutation carcinogenesis model was used to calculate the expected lung cancer incidence caused by both smoking and exposure to radon in two populations, i.e. those of the Netherlands and Sweden. The model parameters were taken from a previous analysis of lung cancer in smokers and uranium miners and the model was applied to the two populations taking into account the smoking habits and exposure to radon. For both countries, the smoking histories and indoor radon exposure data for the period 1910-1995 were reconstructed and used in the calculations. Compared with the number of lung cancer cases observed in 1995 among both males and females in the two countries, the calculations show that between 72% and 94% of the registered lung cancer cases may be attributable to the combined effects of radon and smoking. In the Netherlands, a portion of about 4% and in Sweden, a portion of about 20% of the lung cancer cases (at ages 0-80 years) may be attributable to radon exposure, the numbers for males being slightly lower than for females. In the Netherlands, the proportions of lung cancers attributable to smoking are 91% for males and 71% for females; in Sweden, the figures are 70% and 56%, respectively. The risk from radon exposure is dependent on gender and cigarette smoking: the excess absolute risk for continuous exposure to 100 Bq m-3 ranges between 0.003 and 0.006 and compares well with current estimates, e.g. 0.0043 of the International Commission on Radiological Protection (ICRP). The excess relative risk for continuous exposure to 100 Bq m-3 shows a larger variation, ranging generally between 0.1 for smokers and 1.0 for non-smokers. The results support the assumption that exposure to (indoor) radon, even at a level as low as background radiation, causes lung cancer proportional to the dose and is consistent with risk factors derived from the miners data.     

Extrapolations are the Achilles heel of risk assessment

This Reflections article considers the problems associated with the various extrapolations that are required for the estimation of human cancer risks from exposure to environmental carcinogens at low doses. These include extrapolation between species (particularly rodent to human), from responses at high doses to those at low doses, and among different stages of life. Reductions in uncertainty in risk estimates are closely coupled to the ability to conduct reliable extrapolations. The best way forward appears to be the use of data on mechanisms of carcinogenesis to develop bioindicators of responses related to the pathway to tumor formation. Such an approach is proposed based on the phenotypes represented by the six acquired characteristics forming the Hanahan-Weinberg model for carcinogenesis (The Hallmarks of Cancer). In addition, approaches can be established that use the Hanahan-Weinberg model as the basis for the collection and/or analysis of microarray or similar data. The reduction in reliance on default options and safety factors in the risk assessment process is a real possibility.