Abstract
The classification of harmful radiation-induced effects into ‘stochastic’ and ‘deterministic’ or ‘cancer/heritable effects’ and ‘tissue reactions’ has played a significant role in the recommendations of the International Commission on Radiological Protection (ICRP) over a number of decades. More recently, however, there has been a debate regarding the adequacy of the current scheme for classification of health effects with current scientific evidence. While these scientific aspects are discussed in greater detail in a parallel article, we will emphasise the practical importance of the classification itself here. The setting of dose limits is quite different for ‘deterministic’ and ‘stochastic’ effects. In the first case, provided reliable data regarding the threshold doses for radiation effects on a particular tissue are available, the dose limit is supposed to avoid harm altogether. With ‘stochastic effects’, the expected risks at the dose limit are supposed to be ‘tolerable’. Below the dose limit, optimisation must be aimed for based on the ‘prudent’ assumption that the stochastic risks linearly depend on the dose and that there is no threshold. If the distinction between the two kinds of harmful effects is drawn into question, possible consequences for the system of radiological protection need to be considered, especially in the context of its aims as they currently apply, namely, ‘to manage and control exposures to ionising radiation so that deterministic effects are prevented, and the risks of stochastic effects are reduced to the extent reasonably achievable’. It may be necessary, moreover, to discuss if and how the terminology used by ICRP (and others), in particular the terms ‘deterministic effects’ and ‘tissue reactions’, should continue to be used for the classification of radiation effects.
INTRODUCTION
The classification of harmful radiation-induced effects into ‘stochastic’ and ‘deterministic’ or ‘cancer/heritable effects’ and ‘tissue reactions’ goes back a number of decades. Until the 1950s, the International Commission on Radiological Protection (ICRP) took it for granted that the most important radiation effects on human health, such as skin reactions and suppression of haematopoiesis, occurred only if a specific ‘threshold dose’ was exceeded. Accumulating evidence from animal experiments then suggested that germline mutations were induced proportionally to dose without a threshold. The 1956/1957 amendment to the ICRP recommendation of 1954 (ICRP, 1958) stated that ‘it is prudent to limit the dose of radiation received by gametes’. In due course, the ‘linear non-threshold model’ was also found to be most applicable and recommendable for the induction of cancer (ICRP, 1966). Cancer and heritable effects were termed ‘stochastic’ because of the probabilistic nature of their occurrence. The effects induced with a threshold dose were accordingly called ‘non-stochastic’ until 1990 (ICRP, 1991) when the term ‘deterministic’ was introduced. It is still used in the general recommendations of 2007 (ICRP, 2007). However, ICRP has also pointed out that because these effects are modifiable by different factors and thus not entirely predetermined, the ‘directly descriptive term ‘tissue reactions’ might be preferable.
Here, we will not discuss which effects for which reason should be termed ‘stochastic’ or ‘deterministic’. To some extent, that is addressed in a parallel article (Ainsbury et al., 2024). We will instead emphasise the practical importance of the classification itself. For instance, the rationale for dose limits differs for ‘stochastic’ and ‘deterministic’ effects. With ‘stochastic effects’, as mentioned above, the keyword is ‘prudence’: The linear non-threshold model is assumed to apply, although direct evidence below a few tens of mSv is currently scarce. The expected risks at the dose limit should at least be ‘tolerable’. Moreover, optimisation must be aimed for below the dose limit: ‘Exposures should be kept as low as reasonably achievable, taking into account economic and societal factors’, which again is based on the ‘prudent’ assumption that there is no threshold dose. Similar considerations apply to all exposure situations and exposure categories, i.e. not only for dose limits, but for constraints and reference values as well.
With ‘deterministic effects’, provided reliable data regarding the threshold doses for radiation effects on particular tissues are available, the setting of dose limits would seem to be more straightforward: They are supposed to avoid harm altogether. However, what has been sometimes overlooked is that there is always a distribution of sensitivities within a population, so absolute safety cannot be guaranteed. ICRP Publication 103 (ICRP, 2007) suggested defining the ‘threshold dose’ as the ‘dose estimated to result in only 1% incidence of tissue reactions’. In radiotherapy, it may sometimes be impossible to keep tissue reactions at such a low level. Even here, however, doses must be optimised in order to reduce severity, and adequate measures must be taken to mitigate the consequences of high doses and subsequent tissue reactions.
RELEVANT STATEMENTS OF ICRP
In the latest general recommendations of ICRP (2007), the distinction between two kinds of harmful radiation-induced effects plays a central role in the ‘Aims of the Recommendations’: (29) The Commission's system of radiological protection aims primarily to protect human health. Its health objectives are relatively straightforward: to manage and control exposures to ionising radiation so that deterministic effects are prevented, and the risks of stochastic effects are reduced to the extent reasonably achievable.
Stochastic effects are ‘Malignant disease and heritable effects for which the probability of an effect occurring, but not its severity, is regarded as a function of dose without threshold’. Not in the glossary, but in Annex A (A54), it is stated that ‘Effects resulting from damage in a single cell are termed stochastic effects’.
ICRP does take a look at the history of these terms, quoting some of its earlier publications to see how the terminology regarding the two types of harmful radiation effects developed over the years: (56) In Publication 60 (ICRP, 1991b) the Commission classified the radiation effects that result in tissue reactions as deterministic effects and used the term stochastic effects for radiation-induced cancer and heritable disease. Effects caused by injury in populations of cells were called nonstochastic in Publication 41 (ICRP, 1984), and this was replaced by the term deterministic, meaning ‘causally determined by preceding events’ in Publication 60 (ICRP, 1991b). The generic terms, deterministic and stochastic effects, are not always familiar to those outside the field of radiological protection. For this and other reasons (given in Annex A), Chapter 3 and Annex A also use the directly descriptive terms tissue reactions and cancer/heritable effects respectively. However, the Commission recognises that the generic terms, deterministic and stochastic effects, have a firmly embedded use in its system of protection and will use the generic and directly descriptive terms synonymously, according to context.
As to the reasons why ‘deterministic’ might not be an ideal designation but continues to be used, ICRP gives the following reasons: (A 33) An increasing number of studies on early tissue reactions have shown the ability to modify these using various cytokines and growth factors, primarily to stimulate regeneration of progenitor cells. Other biological response modifiers can be used for late reactions, in particular vascular modifying agents that delay the expression of organ damage induced in experimental animal systems. This ability to modify the response of tissues and organs means that the term ‘deterministic effects’ is not entirely accurate because, quantitatively, the effects are not necessarily pre-determined. Nevertheless, this term has become widely and firmly established, and the Commission continues to use the expression ‘deterministic effects’ to denote tissue and organ reactions. (249) In addition to the limits on effective dose, limits were set in Publication 60 for the lens of the eye and localised areas of skin because these tissues will not necessarily be protected against tissue reactions by the limit on effective dose. The relevant values were set out in terms of the equivalent dose. These dose limits remain unchanged .... However, new data on the radiosensitivity of the eye with regard to visual impairment are expected. The Commission will consider these data and their possible significance for the equivalent dose limit for the lens of the eye when they become available. (o) This ICRP report has some changes to indicated threshold doses for tissue reactions compared with those stated in Publication 103 (ICRP, 2007). First, the threshold dose for radiation-induced eye cataracts is now considered to be approximately 0.5 Gy for both acute and fractionated exposures, in line with various recent epidemiological studies. Second, circulatory disease has been recognised as an important late effect of radiation exposure, both for mortality and morbidity. An approximate threshold dose of 0.5 Gy has been proposed for acute and fractionated/protracted exposures on the basis that this might lead to an incidence of the order of 1% of circulatory disease in exposed individuals.
It might be noted that the effective dose limits, which are set and applied with the aim of ensuring that stochastic effects are within the range of tolerability, have not undergone any change since ICRP Publication 60 (ICRP, 1991) and that ICRP has continued recommending the same limits in ICRP Publication 103 (ICRP, 2007) without revisiting the arguments justifying their choice two decades earlier (cf. Zölzer, 2022). This is probably related to the fact that the risk estimates for cancer induction have not changed substantially (ICRP, 2007).
There are also other areas of radiological protection where things are made more apparent or more manageable by the distinction between ‘stochastic’ and ‘deterministic’ effects or ‘cancer/heritable effects’ and ‘tissue reactions’. For instance, the radiation weighting factor wR is based on scientific evidence for cancer induction and heritable effects. The numerical values of wR (and thus the concept of organ equivalent dose) only apply for the protection against ‘stochastic effects’. In contrast, ICRP has recommended that for the assessment of ‘tissue reactions’, values of the relative biological effectiveness (RBE) should be used (ICRP, 2007). This is just an example of the necessity to consider the implications of any change in the classical classification of radiation effects for other parts of the system.
CONSIDERATIONS FOR THE FUTURE EVOLUTION OF THE SYSTEM
ICRP has appointed Task Group 123 to review the ‘Classification of Harmful Radiation-induced Effects on Human Health for Radiological Protection Purposes’. It is to ‘clarify the rationale behind the current classification and the primary protection objectives of the ICRP System’ (which motivated the current article, among others). The task group will also ‘assess the reasons calling for an evolution based both on a review of scientific literature and relevance for the radiological protection objectives’ (as outlined in the parallel article of Ainsbury et al., 2024). Furthermore, it must, ‘if any evolution is deemed desirable from a scientific point of view, assess the impact on practical management of radiological risk with regards to the radiological protection system objective, for both the prevention of harmful tissue reactions and the limitation of stochastic effects’. Discussions along these lines have just started, but the task group has identified several questions that need to be answered concerning the role of the classification of harmful effects for the system of radiological protection. The questions are, of course, only indicative of what needs to be addressed. It is likely that more will arise as the working group reviews the literature and discusses whether and possibly what changes need to be made to the classification:
- Does the current classification of radiation-induced health effects accurately reflect the best available scientific evidence? - Is it really possible to unambiguously distinguish effects which are characterised by a threshold dose from those more appropriately described by the linear non-threshold model? - Can we safely assume that the effects with threshold are due to ‘injury in populations of normal cells’, while those without threshold result from ‘damage in a single cell’? - Is it always the severity of the effect that is dose-dependent with population effects and the probability with single cell effects, or can both severity and probability be affected? - Should we take into consideration not only the dose dependence, the mechanism of action, and the effect measure but also the timeline (early vs late effects, acute vs chronic effects)? - Can the differences in all these aspects be captured with just two categories? - What are the practical implications of a (possible) change in the classification of radiation-induced health effects? - If we let go of the ‘threshold’ vs ‘non-threshold’ distinction, what would that mean for the general aim of radiological protection to prevent deterministic effects and reduce the risks of stochastic effects to the extent reasonably achievable (ICRP, 2007)? - If the current classification is to be maintained, should we continue to use both generic terms (‘deterministic vs stochastic effects’) and descriptive terms (‘tissue reactions vs cancer/heritable effects’), and even mix them at times (‘tissue reactions vs stochastic effects’)? Would there be merit in returning to ‘non-stochastic vs stochastic effects’ (ICRP, 1984)? - Whether or not the current classification is changed, we need to be clear about the criteria or critical considerations which allow us to distinguish between one category and the others.