Biological effects of radiation in combination with other
physical, chemical or biological agents 

Annex L from Ionizing Radiation: Sources and Biological Effects 

United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 1982 Report to the General Assembly

United Nations, New York, 1982

[ Index | Next page | References ]

[Mindfully.org note: This first section (paragraphs 222-235) is from the book's Introduction. Following it is Annex L]

4. Biological effects of radiation in combination with other agents¹⁶

16 This subject is reviewed extensively in Annex L "Biological effects of radiation in combination with other physical, chemical and biological agents".

222. The combined effects of radiation and of other physical, chemical and biological agents are potentially of great importance but the relevant data are scattered and inconsistent. Therefore the emphasis of this review has been mainly theoretical, with illustrative examples of the complexities of the subject drawn from experimental and epidemiological reports. Except for the case of tobacco smoke, which may act synergistically with radiation in producing lung cancers under some working conditions, this study has been unable to document in man any clear case of interaction, at least of the kind which may result in substantial modifications of the estimates of risk for significant sections of the population. The Committee has outlined the main directions along which future work might be usefully pursued since data on combined effects are at present inadequate.
223. The joint effects of ionizing radiation and other physical, chemical or biological agents are of potentially great importance because radiation is ubiquitous in nature and in modern life many situations could be envisaged which might lead to some form of interaction.
224. In spite of many reports claiming or showing some kind of interaction, the Committee believes that the results of these studies are, on the whole, inconclusive, for a number of reasons. First, when considered comprehensively in the light of the Committee's objectives, these reports appeared to involve exposure levels much higher than the environmental levels of practical significance, and to involve single, rather than protracted, exposures. Secondly, there was a lack of any systematic treatment of each case of interaction in regard to the dosage of the interacting agents and to the interaction mechanisms. Thirdly, many of the reports made little use of appropriate methodologies of analysis, although these had long been available in other fields of the biological sciences. Finally, the absence of sound conceptual bases about the possible nature of the interaction made it impossible to define this notion to even a moderate degree of refinement.
225. Given the above situation, the Committee assumed that a preliminary theoretical treatment of the field in an attempt to suggest definitions, to identify methodologies of analysis, and to exemplify the complex nature of the problems with practical examples, would be more appropriate than a systematic review of literature reports. The Committee considered two possible types of interactions. In the first type both ionizing radiation and the other interacting agent may each produce some effect: here, additivity, synergism and antagonism are seen as the three possible conditions of interaction. The second type is that between ionizing radiation and any other agent which is by itself inactive when administered alone: protection and sensitization are here the terms describing the reduction or the enhancement, respectively, of the effects of radiation acting alone. Such a classification is not an absolute one because the doses of the interacting agents and the types of effect may influence profoundly the nature and degree of the interaction. Cancer-promoting substances were examined as a special case.
226. The concepts of exposure, dose and response as applicable to the special case of combined actions were first discussed. The Committee then reviewed the existing methodologies of analysis, which might allow an assessment, at least qualitative, of the results of combined treatments. A more detailed probabilistic discussion of this subject was also provided leading, under certain conditions, to a precise description of the interaction factors. Attention was given to the applicability of these basic but rather abstract concepts to practical situations in the presence of complex biological effects.
227. In order to produce meaningful answers, the biological effects under study must be well defined and explored for the full range of doses of the interacting agents, applied both separately and jointly. The temporal pattern of the exposure (contemporaneous or sequential, single or fractionated) and the order of administration of the agents are often of decisive importance in respect to the production of a given type and degree of effect. A detailed knowledge of the mechanisms is also a prerequisite for the assessment of the conditions and the level of interaction. However, in much of the work examined these basic requirements were not met or were only imperfectly explored; also, the statistical significance of the results was often so low as to make any assessment of interaction at best suggestive.
228. Regarding the interaction of radiation and other physical agents, the available information was mostly on interactions between different types of ionizing radiation or between ionizing radiation, on the one hand, and ultraviolet radiation, microwaves and heat, on the other. Some synergistic action was apparently reported in workers in the radiotechnical industry exposed jointly to ionizing radiation and microwaves. Functional disturbances of the autonomic nervous system and subjective symptoms of discomfort were the effects under study. A critical analysis of the data showed that the nature of the symptoms, the difficulty with their quantification, the insufficiently controlled conditions of exposure and the incomplete statistics were all reasons to regard these reports with some reservation. Fewer data were available on the combined action of radiation with high altitude, physical stress, mechanical damage and ultrasound, and the results seemed on the whole inconclusive.
229. Many different classes of chemical compounds have been examined for their possible interaction with radiation. Inorganic compounds containing lead, cadmium, chlorine, beryllium and platinum may be of importance under special conditions of work and the very limited experience available could profitably be enlarged for more definitive conclusions. Data on various types of dust were thought to be very uncertain because additive, synergistic and inhibitory effects were described, to a degree not exceeding a factor of four under the worst possible circumstances, compared with the effects induced by radiation alone. Antibiotics, chemotherapeutic substances and other pharmacological agents appeared to be of more significance under special clinical situations than for the population at large.
230. The possible combined action of radiation with compounds known for their carcinogenic properties was the object of special attention. Although the information reviewed concerned a variety of initiators and promoters, the data available for each of these substances were very incomplete and the evidence conflicting. No final statement could be offered in regard to any substance or to any class of tumour unless the dose, the dosage schedule and the treatment modalities of the combined treatments had been analysed to a greater depth. The experience on benzo(a)pyrene, diethylnitrosamine, various types of dust and oil exhaust fumes might be enlarged for firmer conclusions, in view of the widespread environmental presence of these substances.
231. It appears that in man tobacco smoke may act by shortening the time of appearance of lung cancer induced by alpha particles of radon daughters. It is not yet clear whether such an action might result from promotion by some specific component of tobacco smoke, or might be ascribed to other nonspecific effects on the respiratory tissues. The precise evaluation of the interaction factor may depend critically on the length of the observation period, as well as on the age structure and exposure history of the persons at risk.
232. In animals, there is evidence that some hormones may affect the time or rate of appearance of radiation-induced tumours, particularly of the mammary gland. This type of synergism is mainly expressed through a shortening of the time necessary for tumour induction. There is, however, a large variability of the synergistic effect with the strain of the animals, such that the same treatment schedule will produce synergism in some strains and antagonism in others. There is also variability in relation to tumour type. In man direct information is lacking. Other biological agents such as viruses and bacteria, or changes in diet, when applied in conjunction with radiation, have produced equivocal or negative results.

5. Summary and conclusions

233. The studies carried out by the Committee in the area of biological effects of ionizing radiation have not resulted in major revisions of the current thinking about the genetic risk estimates or the somatic effects analyzed. They have however focussed on some important new developments and have led to refinements of previous knowledge. On the whole, these new studies have strengthened the Committee's belief that the mechanisms of some radiation effects are becoming reasonably well understood. This applies particularly to non-stochastic effects. 

234. For other effects, such as those depending on the neoplastic transformation of the irradiated cells, present knowledge of mechanisms is still largely incomplete. A further analysis of cancer induction mechanisms will be undertaken when the dosimetry in Hiroshima and Nagasaki survivors is clarified. The Committee will continue its surveillance and reviewing of the whole field of radiation carcinogenesis, including the theoretical foundations and the actual risk estimates of cancer induction in man.

235. With regard to hereditary effects, the Committee notes that further advances have been made in our knowledge of the dose-response kinetics and other aspects of some of the more important types of genetic change which can be induced by radiation in experimental mammals. Extensive use of experimental data for genetic risk assessment is still considered essential in the absence of significant positive results with respect to hereditary effects after human exposures. A new method has been developed for assessing the magnitude of first-generation risks from harmful dominant mutations. This approach and other methods for estimating genetic risks in the progeny of those exposed to low radiation doses have yielded very similar results. However, many important problems remain. For instance, human female germ-cells are considered to be less sensitive than male ones for the induction of genetic damage from low-level radiation, but the actual magnitude of this difference is still uncertain. Further work will also be needed on the extent to which recessive mutations lead to genetic damage over many generations after the first. However, advances in human genetics and new methods of comparing mutation rates in human and animal cells should help to solve some of these outstanding problems.

ANNEX L 
Biological effects of radiation in combination with other physical, chemical or biological agents

1. ANNEX L   CONTENTS Paragraphs INTRODUCTION 1-20 I. MODES OF INTERACTION 21-72 General approach 21-41 Surface of response and isobolic diagrams 42-49 Probabilistic assessment of the interaction 50-61 Theory and practice 62-72 II. PHYSICAL AGENTS 73-113 A. Combinations of various types of ionizing radiation 73-78 B. UV and ionizing radiation 79-85 C. Electromagnetic and ionizing radiation 86-93 Experimental data 86-91 Epidemiological evidence 92-93 D. Suboptimal temperature and ionizing radiation 94-103 High temperature 94-99 Low temperature 100-103 E. Magnetic fields and ultrasound 104-107 F. Dusts and fibres 108-113 III. CHEMICAL AGENTS 114-199 A. Inorganic compounds 114-120 B. Organic radiosensitizing compounds 121-136 C. Carcinogenic chemicals 137-157 D. The special case of tobacco smoke 158-183 General 158-159 Experimental data 160-168 Epidemiological evidence 169-183 E. Other drugs 184-199 IV. BIOLOGICAL AGENTS 200-217 General 200-201 Hormones 202-213 Infectious agents 214-217 Viral infections 214-215 Bacterial infections 216-217 V. CONCLUSIONS 218-237 VI. RESEARCH NEEDS 238-244 Page References 765

   In man's living and working environments situations are often encountered in which different ambient factors of a physical, chemical or biological nature could conceivably combine with ionizing radiation in giving rise to undesirable effects. In this paper for the first time the Committee considers the combined action of radiation with potentially important environmental conditions. Since this paper concentrates on radiation in environmental circumstances, three important areas of combined action between radiation and chemical agents are not considered here. The first concerns the combined action of chemical agents (both chemotherapeutic compounds and sensitizers of various kinds) to enhance radiation effects in clinical radiotherapy [B28, D24, D25, H28, H29]. The second results from the restriction of this paper to radiation effects combined with agents which affect carcinogenesis, not therefore including combined effects in mutagenesis. This area may be considered by the Committee in the future. The third area not treated in detail in this paper is the effects of a combination of protective agents with acute radiation exposure [A12] because this subject is of only minor importance in estimation or modification of risk. 

2. There is a great scarcity of systematic data on which an analysis of combined effects can be based, in spite of the large number of reports where combined actions were tested and interactions claimed. Thus, this Annex must be somewhat different from others in which a large body of literature data is reviewed and systematically analysed. This Annex will be instead more hypothetical and will attempt to suggest definitions, to identify suitable methods of analysis, to select from a large amount of diffuse information the conditions and the data of importance for further consideration and to provide suggestions for future research.

3. The following review of experimental or epidemiological data should be simply taken as an illustration of some theoretical analyses using examples from the literature. These considerations point, on the one hand, to the very preliminary character of this Annex and, on the other hand, express a word of caution against hasty conclusions in view of the present state of knowledge and the large variety of situations encountered.

4. There are many instances of possible combined actions in which different agents may interact with ionizing radiation. Among the physical agents, for example, temperature should be considered. It is well known that ambient temperature in different environments may vary within a range of about 70°C although control mechanisms allow man to survive under the most extreme conditions. It is also known, however, that small changes in the temperature of cells may result in striking changes of cell survival upon irradiation. These changes are presently being investigated for their potential in cancer therapy [C1, F2, D16]. Ultraviolet light, itself carcinogenic, sound, ultrasound and vibrations are present in many living and working environments and may give rise to combined actions. The same can be said for static electromagnetic fields, and high-frequency or very high frequency (microwaves) electromagnetic radiation.
5. Man-made (xenobiotic) chemicals in the environment are a major concern for toxicologists. According to some estimates [M2] the number of identified molecules is now more than four million and every year a few hundred thousand new items are added to the list. There are some tens of thousand chemicals in common use in modern societies, not including pesticides, pharmaceuticals and food additives. The so-called "energy related pollutants" are also to be considered in this category. They include the oxydes of carbon and nitrogen, sulfur compounds, polycyclic hydrocarbons and some others. Among them 3,4-benzo(a)pyrene (BP) is frequently used as an index of polycyclic hydrocarbons with carcinogenic properties [S2]. Its yearly production is estimated to be approximately 5000 t [S31]. The time course of its production may be followed, for example, by lake sediment analysis [H11]. The concentration of BP in the air of large industrial cities may reach values of 100 ng/ m³ [BIO]. BP is also one of the many chemical constituents of tobacco smoke and may be considered of importance for some sections of the population occupationally exposed to radiation. The circulation of BP and of other polycyclic aromatic compounds in the environment has been studied extensively [S2, S31].
6. The list of chemicals whose action might combine with that of radiation in the environment is very extensive. Special attention should be given to situations of practical interest where the chemical agents themselves have carcinogenic properties [H21]. For example, many industrial effluents contain trace elements such as arsenic, nickel or chromium. These substances may produce carcinogenic or mutagenic effects [T7]. The same is true for dust and fibres. Dust is a very common and widespread industrial emission and a component of many occupational environments. It has been reported that dust or fly ash from power stations may have carcinogenic properties [K9] or may serve as carriers of trace metals, radioactive nuclides or polycyclic aromatic hydrocarbons [B22]. In mines mineral dust may combine with the organic products of diesel exhausts and with radioactive radon and thoron daughters [C16]. Asbestos fibres are also often a significant component of occupational and home environments which may include ionizing radiation.
7. High levels of mutagenic chemicals have been reported in many types of food [S32]. Broiled meat and fish contain mutagenic compounds arising from the pyrolysis of proteins and aminoacids. Mutagens and co-mutagens have also been reported in derivatives of vegetable foods, such as caffeine. As mutagenicity often correlates well with carcinogenicity, the above substances may be considered potential carcinogens, both alone or in combination with radiation. According to some estimates [H2] up to 20–50% of spontaneously occurring human tumours can be attributed to diet. Some pharmaceutical substances are also known for their carcinogenic potential: depending on their use and diffusion they could also be considered as candidates for combined actions.
8. Among biological agents, viruses may be regarded as environmental factors likely to interact with radiation. It is well known that some viruses have an important role in the aetiology of some radiation-induced animal tumours as specific agents. There is a possibility that specific agents of a similar nature may be involved in the induction of tumours in the human species and non-specific associations or combined actions, even though on a purely speculative basis, may be visualized. Natural hormones could also be viewed as a special case of interaction in view of the well-known dependence on the hormone level of some forms of radiation-induced tumours in experimental animals.
9. There are two ways of carrying out an analysis of combined actions. The first is to search for any possible effect, whatever its practical significance or quantitative value might be. The second is to concentrate on those effects that may be of importance for the assessment of risk in man. The first approach is that to be followed in the present preliminary analysis. The present practice in radiation protection is that of assuming sensitivity values across the population which apply to all groups, e.g., to males and females of all ages. This practice does not deny the existence of real changes in the susceptibility between various classes of people, but recognizes the convenience that for practical purposes a single average value of the risk is desirable and sufficient.
10. In acknowledging the merits of this approach, the Committee wishes to emphasize that unless the effects to be validated as synergistic or antagonistic are extremely important (i.e., unless they might lead to changes of at least an order of magnitude in the risk estimates) and unless they also applied to substantial fractions of the population at large, they presumably may not be of relevance in assessing risk estimates in man. The above consideration applies to the estimation of risks for radiation protection purposes. It does not contradict the fact that if some synergistic or antagonistic effects can be identified under specific exposure conditions of occupational or medical relevance, appropriate actions should be taken to change such conditions. Under such circumstances, however, the problem would not be any longer one of radiation protection philosophy, but rather one of practical occupational medicine. It would not involve basic changes in the approach to such matters but specific remedial local actions.
11. Radiation effects with particular regard to carcinogenic and to genetic and developmental consequences of irradiation were considered by the Committee in its 1977 report [U1]. Non-stochastic effects of whole- or partial-body irradiation (Annexes K and 3, respectively) and genetic effects (Annex I) are also discussed in this report. When reviewing such a broad field as that of combined actions, no effects should be excluded from consideration at whatever level (subcellular, cellular, tissue, organ, whole-body) they may be manifested. This is particularly true in view of the heterogeneity of the data available and of the fact that understanding of combined effects will eventually require knowledge of the mechanisms involved. That is why effects other than those mentioned above will be discussed in this Annex. However, the main emphasis will be on stochastic effects. Where possible, epidemiological data will be considered, even though studies of this sort are rare and often statistically inconclusive.
12. Each of the possible interacting agents may act alone in producing biological effects or may only be active in conjunction with other factors, particularly radiation. Exposure to any of these agents may be acute, subacute or chronic, within a wide range of doses and dosages. The pattern of exposure may also play a role, as the contemporaneous action of the various agents or the order of their sequence and the intervals between treatments may conceivably affect the quality or the degree of the effect. Of all possible situations of combined actions the Committee chose to particularly investigate conditions where long-term exposure to low levels of the agents on large human populations may apply, because these conditions may possibly affect radiation risk estimates in man.
13. The combined action of several agents is not a new problem in medicine. As early as 1928 Loewe [L1] quantitatively reviewed the approaches to the assessment of the action of combined drugs. So-called "isobolic diagrams" were proposed in this regard. This Annex will consider this approach in detail, as well as other approaches extensively used by toxicologists [M1, T1]. Some of these ideas were adapted specifically for the needs of the Annex and illustrative material has also been derived and modified for the same purpose.
14. Nomenclature in the analysis of combined actions was a problem that was recognized very early [L1]. In order to simplify the discussion to follow, it is appropriate to provide some clear definitions and terminology. Two classes of combined effects will be considered. In the first class, both ionizing radiation and the other agent (or agents) produce the effect under discussion. The second class includes the combinations where ionizing radiation produces an effect whose nature or amount may be modified by the other agent which by itself is inactive. This classification is only made as a convenient approximation.
15. For the first class of interaction there are three types of combined actions. When the end-effect of the combined action equals the sum of effects of the two agents acting independently, the resulting situation is one of "additivity". If additivity does not apply, then there are two possibilities. When the effect of the combined action exceeds the sum of the effects produced separately by the agents, the situation is one of "synergism". Finally, when the combined action results in an effect which is less than expected from the sum of the action of the interacting agents, the situation is termed "antagonism". The precise meaning of the "sum of effects" will be expanded further in chapter I. The notion of summation of effects pre-supposes the existence of a quantity which may be meaningfully added.
16. The concept of additivity cannot be extended to the second class of combinations since radiation is here the only agent capable of producing an effect. Under these circumstances the comparison is usually between doses of radiation producing the same amount of effect in the absence or in the presence of the modifying agent. If, for a certain degree of effect, the dose of radiation required is greater in the presence of the modifying agent, the resulting action is termed "protection". Conversely, when the dose of radiation is less for the same degree of effect in the presence of a modifying agent, "sensitization" occurs.
17. The above classification is not an absolute one. For example, sensitizing substances which have been assumed to be inactive, may be able to produce some effect at high exposure levels. Also, if one considers carcinogenesis as an effect, promoters may be viewed as a special case of sensitizers and many promoters may show initiating properties. The low environmental levels of the interacting agents are mainly those of interest in this Annex. At these levels the threshold-type dose-response curves of the sensitizers and promoters may render their contribution negligible or zero. A unified approach to both classes of interaction in terms of interaction coefficient and more precise quantitative definitions of the concepts introduced in the above paragraphs will be developed in chapter I.
18. For exposure of the public the most significant man-made source of irradiation is for diagnostic medical purposes where the yearly dose equivalents may be up to the order of a few millisievert (mSv) (see Annex G). Sources of occupational exposure are much more varied and may range from exposure to radon in mines to x rays generated by electronic appliances. The yearly occupational exposure according to ICRP recommendations should not exceed 50 mSv [Il]. Average yearly exposures to natural sources of radiation are between 2 and 3 mSv (see Annex B). The actual occupational exposure in industry has average values of about 5 mSv (see Annex H). Thus the other physical, chemical or biological environmental agents would combine with ionizing radiation at levels of the latter of 1—10 mSv per year. These levels are usually referred to as low doses.
19. It is sometimes held that in view of the ubiquitous nature of background radiation all experimental or epidemiological studies on the toxicity, carcinogenicity or mutagenicity of chemicals or other agents are automatically performed to account for the concomitant radiation risk. All the relevant risk assessments would therefore be in essence assessments of combined action [Si]. This may be too broad a generalization for the following reasons. Firstly, the actual levels of exposure to ionizing radiation may be orders of magnitude higher than those cited in the preceding paragraph, and the levels of the other agents orders of magnitude lower than those at which experimental risk assessments were performed. In view of the non-linearity of the dose-response relationships for most chemical agents, extrapolation of the risk assessments between such widely different situations would be unwarranted. On the other hand, some chemicals which are ineffective in producing detrimental changes when acting alone, may instead provide a significant modification of the radiation action, as in the case of carcinogenic promoting substances. Animal experiments are usually carried out at levels of exposure to chemical or other agents which are much higher than those found in the environment, which weakens the basis for extrapolation. Under such conditions of great uncertainty the best course of action is to reserve any judgement and to investigate the facts.
20. In summary, the scope of this Annex is:

1. To review possible quantitative approaches to the assessment of the combined action of radiation and other environmental conditions, based on the concepts of additivity, synergism, antagonism, sensitization and protection;
2. To explore whether and to what extent concepts in other fields of the biological sciences may be applied to the special case of interaction with radiations, particularly at very low doses of the combining agents;
3. To consider experimental results on the combined action of radiation and other conditions, in order to elucidate possible mechanisms of action that may allow generalizations and extrapolations;
4. To review existing epidemiological data on subgroups of populations living or working under the action of radiation and other environmental toxic agents;
5. To identify possible areas for useful research in the field of combined effects.

I. MODES OF INTERACTION

A. GENERAL APPROACH

21. When examining the concept of combined action it is useful to start with the definition of a quantity referred to here as "exposure", X, which may apply to any environmental agent [L3, L4]. Exposure is the independent variable in exposure-effect relationships. Without exposure to the agent there can be no effect over the spontaneous level and with increasing exposure the effect appears to follow some kind of functional "exposure-response" relationship. This generalized concept of exposure is different from the notion of exposure in radiation physics (see Annex A). In the case of ionizing radiation the absorbed dose, D, is used instead of the exposure and "dose-response" relationships are established to functionally relate the energy absorbed by the irradiated object with the response observed. If radiation quality must be taken into account, the quantity defined as dose equivalent, H, may be used in place of the generalized concept of exposure, X.

22. The definition of exposure (or dose), X, is more difficult in the case of other agents [L3]. Often this notion includes the product of some intensive quantity (e.g., energy flux per unit area per unit time) multiplied by an extensive quantity (e.g., the time during which the agent acts on the biological system). It has been proposed in the case of chemical compounds [E1] to define exposure as the number of primary chemical events leading to the final effect, but at present the nature of such events is only known in rare cases and their quantification exceptional. The concentration of an agent may often meaningfully be taken as the intensive quantity, multiplied by time as the extensive one. The notion of exposure is by definition extensive and in the case of a chemical substance it could be represented by the formula:
    
                                                                                                         (1)
    
    Obviously, to give exposure some biological meaning, the concentration of the agent, C(t), should be expressed at the level of the target biological structure, but this is often impossible. A useful type of exposure characterization such as the pharmacological dose (the quantity of the chemical introduced per unit weight of the organism) does not provide such information. In these cases special assumptions concerning the intake, retention, metabolism and excretion of the agent under investigation must be made [L3, W4].

23. Even the relatively simple case of a chemical acting on a culture of cells in vitro may require special consideration of the kinetics of the substances involved and of the different forms of their possible transformation [W4]. One may, for example, consider a scheme whereby a chemical A is converted into intermediate B which is in turn transformed into a cell-bound moiety C leading to the observed effect:
    
                                                                                           
    
    Clearly the concentration of C is the quantity to be used in equation (1) to express the exposure. It often happens, however, that the only information available is on the chemical A, the most readily measurable quantity, and this information may not be directly proportional to the values for C. Thus, there might be apparent absence of effect, in spite of a high concentration of A, on account of absence of moiety C, at least at the beginning of exposure.

24. Thus, the metabolic activation of chemicals into active forms is of great interest [M19, S33]. Chemical carcinogens are known to be subject to complex processes of enzymatic reactions in vivo. The chemical compound introduced into the body may be considered as a pre-carcinogen which, through various reaction pathways, will eventually produce proximate and ultimate carcinogenic derivatives. From a purely chemical point of view, one of the important generalizations of the recent years is that the ultimate forms of chemical carcinogens are usually electrophilic (i.e., electron-deficient) reactants. Many specialized examples of such processes are considered in the above mentioned reviews [M19, S33].

25. In some cases the binding of chemicals with cell constituents may be monitored by the use of radioactive labels. Examples of such studies in vitro with two derivatives of nitrosourea were provided in [W4]. Experiments in vivo are also available [El, W9, P8] in which correlations are established between the administered doses of the compounds, the amount of bound moieties and the biological effects. These studies help clarify the concepts of administered versus active doses of the compounds.
26. If the exposure, X, to a given interacting agent (or to several agents) may be satisfactorily defined, the definition of the effect, Y, should be considered. There are different ways of expressing in quantitative terms the response of a biological object. Y may be, for example, the fraction of cells showing loss of a specific function or the fraction of exposed animals affected by a given mutation or carrying a given type of tumour. In such cases Y describes the probability of induction of that given effect as a result of the exposure X. In other cases Y may describe the degree of a given effect: for example, the weight loss of an exposed animal, the mean number of tumours per animal, changes in various haematological indices. Graded effects may sometimes be reduced to probabilistic quantities by appropriate analysis, but this is not always the case and it may represent a limitation.
27. The simplest functional relationship between exposure and response, Y = F(X), is the linear one:
    
                                                                    Y = Y_o + kX                             (2)
    
    Here the term Yo accounts for the effect produced in the absence of exposure or of any other known cause in an apparently spontaneous fashion. The coefficient k defines the sensitivity of the biological system to the agent. When the separate action of each agent is described by equation (2), then the increment of response of the system to each agent may be written as
    
                                                                     DY=Y — Y₀=kX                   (3)
    
    If one assumes that the increments of response to one agent are independent of the presence of the other interacting agent, the increment of response for the simultaneous action will equal the sum of increments AY1, DY₂
    
                                                                                       DY = k₁X₁ + k₂X₂                          (4)
    
    This is the situation of additivity.
28. However, the experimental value of DY in case of a combined action can be higher or lower than the AY expected from equation (4). If DAYobs > AYexp the situation is defined as synergism. If XXX_AYobs < AYexp the situation is one of antagonism. As a measure of the deviation of the experimental results from additivity one may introduce an interaction factor
    
                                                                      = DY_obs/DY_exp               (5)
    
    The value of = 1 will correspond to additivity, > 1 to synergism and < 1 to antagonism.
29. The above concepts may be represented in a graphical form as in Figure I. Here a given level of response Y* is chosen, which level may be obtained by the action of each agent separately (X*₁ or X*₂, respectively) or by the combined action of both agents at variable exposures X₁ or X₂. If additivity is operating and equation (4) is applicable, all points (X₁, X₂) producing the level of response Y* must lie on the middle diagonal line of Figure I. This line is called the isobolic line and the diagram is called isobolic diagram [L1]. The scale of Figure I is chosen in such a way that the coordinate value equals 1 for each agent acting separately, that is, X₁/X*₁ = 1 and X₂/X*₂ = 1.

 

Figure I. Isobolic diagram In case of linear additive response to the action of two agents



 

30. The isobolic line in Figure I describes an ideal case of additivity, but all experimental exposure-response relationships are affected by errors. In real situations therefore the line of additivity expands to an area of additivity, such as that covered by the horizontal shading lines in the same figure. If the exposure-response relationships for the agents acting separately are linear, the type of interaction may be defined by simple graphical procedures. For a given level of effect, Y*, several levels of exposure to both agents are tested: if the experimental points (X₁, X₂) fall into the area of additivity, the interaction will be regarded as additive. If the points fall to the left of the area of additivity, the interaction will be one of synergism; and, conversely, one will be dealing with an antagonistic interaction when the experimental points are found on the right-hand side of the area. In Figure I the experimental point A would be regarded as confirming synergism, experimental point B as confirming an antagonistic interaction.
31. As an example of the application of this analysis, the experiments of Murthy et al. [M3] on diploid yeast BZ34 may be of interest. The cells were irradiated by ²¹⁰Po alpha particles or by ⁶⁰Co gamma rays separately or in combination. The end-point studied was reversion to arginine independence. Linear dose-response relationships were found for both radiations given separately with slopes of 25.5 ± 2.6 and 10.9 ± 0.4 reversions per 10⁶ survivors per Gy applying to the alpha and to the gamma radiation, respectively. In the case of combined simultaneous treatment with both radiations (25% of the dose was by alpha radiation at 0.5 Gy/min and 75% by gamma radiation at 1.54 Gy/min) the slope of the regression line changed to 17.7 ± 0.9 reversions per 10⁶ survivors per Gy. The results may be interpreted by an isobolic diagram, as in Figure II. For the level of reversion Y* = 180 rev/10⁶ survivors the dose of ⁶⁰Co gamma would be 15 Gy and that of ²¹⁰Po alpha 6.4 Gy. The dashed lines parallel to the isobolic line in Figure II establish the 95% confidence limits. If one plots the points corresponding to the same level of reversions for the two agents combined, one finds the point denoted A which lies clearly to the left of the area of additivity. It is concluded that synergistic interaction of the two agents applies in this case. This is an example of isobolic diagram analysis in its most simple form.

32. In the above example the mutation frequencies could be meaningfully added because their increase with dose was linear. The same procedure is not applicable when the effects change as exponential or sigmoid functions of the dose, unless the dose-response relation-ships may be converted to linear or quasilinear functions.
33. The process of addition itself may be performed in two ways. The first, takes the response to the dose A from the survival curve A and adds it arithmetically to the response to dose B from survival curve B. Both doses are counted from the origin of the co-ordinates. Loewe [L1] designates this type of addition as heteroaddition. A second process of addition, called isoaddition, is also possible. Let us assume that agent A is applied before agent B (Figure III b, e, h). Figure IIIa shows the case when the dose Ao is given before B_o. In the case of heteroaddition the dose B_o. would be counted from the origin of the co-ordinates. In the case of isoaddition, on the contrary, the latter dose would be counted from point O', corresponding to the survival level on curve B to which the biological system is brought by the action of agent A. It is easily appreciated that for isoaddition the response to B_o will be much greater than in the case of heteroaddition. This is the reason why the isobolic lines of iso- and hetero-addition are so different in Figure IIIb.

34. The area between two isobolic lines may be called the envelope of additivity [S3]. As a result of different sequencing of the agents this envelope may reduce to a line [R7], as shown in Figure IIIc. This occurs when one of the two interacting agents produces an exponential response. Other examples (Figure III d, e, f, g, h, k) show how the form of the response curves and their relative curvature define the form of the envelopes of additivity and the influence of a different sequence of the agents.
35. The above considerations may be generalized to any type of exposure-response relationship. Since any a priori judgement about the type of addition (iso- or hetero-addition) is impossible, both possibilities should be accounted for. The practical usefulness of the envelope of additivity lies in the fact that if the experimental points fall within the envelope, additivity is to be expected. When they fall to the left (point A in Figure IV) synergism is operating; and, conversely, antagonism will be operating if they fall to the right (point B in Figure IV). Enlargement of the envelope due to experimental errors is also shown in the same Figure IV. Attention should be drawn to the fact that although in principle the area of antagonism extends from the upper right-hand border of the additivity envelope to infinity, the straight dashed lines in the figure define the area beyond which the administration of one agent requires application of the other at levels greater than its single exposure level for the same effect. Point C in Figure IV lies in such an area where exposure X₁ requires an exposure X₂ greater than unity.

 

Figure IV. Envelope of additivity and areas of synergism and antagonism



 

36. Discussion has so far been limited to the class of interaction where both agents may produce the effect under study. A large number of agents are however known in radiation biology which may modify the radiation response of the system without being themselves active in determining the effect. These modifying agents are called radioprotectors or radio-sensitizers, without regard to their mechanism of action [M10]. The same approach as that used in the preceding paragraphs for the assessment of the interaction type may also be generalized to the modifiers. However, since only radiation dose-response relationships are considered here, a specific approach to sensitization and protection may be developed.
37. Oxygen is one of the most important modifying agents [DI]. Its action is extremely general at all levels of biological organization in the sense that macromolecular, cellular and tissue systems irradiated under oxygen show an enhanced effect compared to that resulting from the same dose delivered under anoxia. This enhanced effect is often expressed as an oxygen enhancement ratio (OER) defined as 
    
                                                      OER = D_{(non-oxygenated)}/D_(oxygenated)                       (6)
    
    expressing the ratio of doses D under anoxia and under oxygen to obtain a given level of effect. Other similar quantities may be used for the description of the effect of different modifiers. For example, the thermal enhancement ratio (TER) in the case of the combined action of radiation and heat, is:
    
                                                     TER = D_{(standard temperature)}/D_{(enhanced temperature)}     (7)
    
    or the dose reduction factor (DRF) for radioprotectors
    
                                                     DRF = D_(protector)/D_{(no protector)}                                  (8)
    
    For radiosensitizers, the factor in common use is the dose modifying factor (DMF)
    
                                                    DMF = D_{(no sensitizer)}/D_(sensitizer)                                   (9)
    
    This quantity defined for a particular level of response is often referred to as enhancement ratio (ER) or sensitizer enhancement ratio (SER) or dose modifying ratio (DMR).
38. Also for modifying agents one may define the increment of effect in the presence of radiation alone, AY, and the increment in the presence of the modifier DY_M. The concept of an interaction factor may also be introduced, as follows
    
                                                    = DY_M/DY                                                               (10)
    
    When linear relationships apply both in the absence and in the presence of the modifier, the value of the interaction factor will coincide with the value of the dose modifying factor (DMF). In Figure V the line OAB is the response in the absence of the modifier and the line OCD the response in the presence of a sensitizer. In this particular case
    
                                                    = Y_C/Y_A                                                                    (11)
    
    and
    
                                                    DMF = X_B/X_A                                                               (12)
    
    However, as the ratio Y_C/Y_A is equal to X_B/X_A both definitions coincide. In this special case of linearity the values of and of DMF will be independent of the level of exposure, because the straight line is fully defined by only one parameter (the slope at 0 exposure or the response at any specific exposure). In geometrical terms, the ratio Y_D/Y_C in Figure V is the same as Y_C/Y_A.

 

Figure V. Linear exposure-response relationships in the absence (OAB) and in the presence (OCD) of a sensitizing agent



 

39. The circumstances differ of course in cases of non-linear exposure-response relationships that would most probably apply to the vast majority of the situations in practice. Figure VI illustrates one such case where the ratio Y_C/Y_A is not any longer equal to, but is actually much smaller than the ratio X_B/X_A. At exposure level X_B the definition of an enhancement ratio is meaningless because the line through point D parallel to the abscissa will never cross the other response curve OAB, but an interaction factor for a modified response as defined in equation (10) may still be applied. However, the value of will depend on the level of exposure or response.

 

40. The situation is further complicated when the application of a modifier significantly changes the general form of the dose-response relationship. In such cases the use of , that is the use of an enhancement ratio in terms of increment of effects, may not be applicable. The solution requires specifically defining a suitable quantitative measure of the modifying effect under the conditions applying to the experimental situation.
41. The concepts and approaches outlined so far are quite sufficient for a discussion of the available scientific literature on the interaction of different agents with radiation. When possible, in the text to follow the concepts of interaction factor and envelope of additivity on isobolic diagrams will be applied. However, further refinements and generalizations of the concepts outlined may be of some value, as in the two following sections. These sections may however be omitted without significant detriment to the understanding of the experimental material reviewed in the chapters to follow.

B. SURFACE OF RESPONSE AND ISOBOLIC DIAGRAMS

42. The methodology of assessment of effects in combined exposures outlined by Loewe [L1, L5] allows a much broader approach to the problem. If, for the sake of clarity, one assumes only two interacting agents, the response to agent 1 is given by the function F₁(X₁) and that to agent 2 by the function F₂(X₂). The simultaneous action of the two agents will result in some new function F(X₁,X₂). The functions F₁(X₁) and F₂(X₂) describe the response on a plane; the new function F(X₁,X₂) describes the response in a three-dimensional space. This new function is called the surface of response. It may be used for any number of agents, and in these cases it will be described in multi-dimensional space. The concept of a surface of response makes the approach to the assessment of interaction geometrically clear. In this case the comparison is drawn between the surface obtained as a result of addition of responses to single agents (surface of additivity) and the surface of response for the function F(X₁,X₂).
43. Let the functions F₁(X₁) and F₂(X₂) be linear with a simple law of addition operating for simultaneous action. Then one obtains the surface of response (and the surface of additivity at the same time) as the inclined plane in Figure VII. Cross-sections of this plane at different levels of response (in Figure VII at Y = 0.5 and at Y = 1.0) will always produce straight lines which are isobolic lines in the sense of Figure I. By choosing the scales of the coordinate along the X₁ and X2 axes it is possible to adjust the angle of the cross section line with the coordinate plane so as to make it equal to 45°.

 

44. The linearity of the functions F₁(X1) and F₂(X₂) is however by no means a necessary condition for obtaining linear isobolic diagrams. The case of S-shaped functions is considered in Figure VIII. The two functions are represented by the curves in the co-ordinate planes YOX1 and YOX2. The surface of additivity (i.e., the dotted surface in Figure VIII) has now also a changing curvature similar to that of a tense sail, but horizontal planes at levels of response Y = 0.5 and Y = 1.0 still transect this surface by straight lines, so that again the isobolic diagrams are of the same type as in Figure I. According to this graphical representation, synergistic interaction is expressed by a deviation from the surface of additive response nearer to the OY axis. The new synergistic surface of response is presented in Figure IX and it resembles an inflated sail. The cross-section of this surface by a horizontal plane at the level of effect Y = 0.5, for example, produces a curve which is the isobolic diagram of a synergistic interaction. The case of antagonism is exemplified in Figure X, where the antagonistic surface of response is further removed from the OY axis, in the form of a sagging sail. Transection of this surface by an ordinate plane (Y = 0.5) results in a curve with a concavity towards the origin of the co-ordinates, i.e., a curvature in the opposite direction than that of the synergistic action.

Figure VIII. Surface of response in case of additivity for curvilinear functions

Figure IX. Synergistic surface of response

Figure X. Antagonistic surface of response

45. The above interactions may be represented by the isobolic diagrams of Figure XI, where the isobolic lines are the cross sections by a horizontal plane at the level of effect Y = 0.5 of the three surfaces of response in Figures VIII, IX and X. Such comparisons can be made at any level of effect, but in the case of agents present in the environment the levels would generally be low. It is therefore of interest to examine the shape of the surfaces of response around the origin of the co-ordinate axes. It is not uncommon that the form of the surfaces might go from a synergistic type to an additive type in the region of low effects. At different levels of effect the interaction might even change from the synergistic to the antagonistic type, or vice versa.

46. Changing situations of this sort are illustrated in Figure XII, where the form of the surface of response is made clearer through other types of cross sections. In the case shown the planes transacting the surface are diagonal, parallel to the Y axis and with an angle of 45° with respect to the X₁ and X2 co-ordinate axes. The areas of the planes transacted by the surface and covered by the dashed lines show the extent of the difference between the real surface and an ideal surface of regular additivity. The plane nearest to the origin shows a strong antagonistic interaction, but the further the transacting plane is removed from the origin, the less important becomes the antagonism; until, at very high levels of exposure, the interaction becomes synergistic. Cross sections of this type can also help in assessing the mode of interaction and are called "interaction diagrams".

 

47. In Figure XII the levels of exposure are limited by the vertical planes chosen, but the levels of response may change. Such changes of response which depend on the relative contribution of X₁ and X₂ form the interaction diagram as shown in Figure XIII. The lower curve in Figure XIII shows the line of interaction (antagonistic interaction) resulting from the transaction of the surface of response by the vertical plane nearest to the origin of the co-ordinates in Figure XII. The case of additivity is represented by the line parallel to the exposure axis, while the case of synergism is described by the curve with upper convexity. Again, the line of additivity divides the space into two portions: an upper one, where interaction is synergistic and a lower one with an inhibitive type of effect. In essence, this method of analysis relies on the comparison between two surfaces: the actual surface of response and the surface corresponding to the presumed additivity of the effects of the two agents given separately. If the real surface of response is higher than additive, there is synergism. Antagonism would operate in the opposite case.

 

Figure XIII. Interaction diagrams in cases of additivity, synergism and antagonism



 

48. Construction of the surface of additivity is a prerequisite for all comparisons. It has however been discussed already that the addition of responses in complex biological systems represents a problem in itself because the results of iso- or hetero-addition depend on the sequence of the interacting agents. Adding to these uncertainties the experimental errors, turn the surface of additivity into a shell of additivity, corresponding to the envelope of additivity of the bi-dimensional representations. Although the actual comparisons should always be performed in relation to a shell of additivity, it is often more convenient to use the two-dimensional representations under the form of isobolic diagrams (or interaction diagrams). Elaborate methods of analysis have been developed for the interaction of several agents [C21].
49. When all the above assumptions have been dealt with, the comparison of the experimental data with the ideal case of additivity is straightforward conceptually and technically simple. For a given combination of exposures (X₁, X₂) the interaction factor () may be calculated as the ratio of the actual ordinate of response to the ordinate of the additivity surface in the point (X₁, X₂). To this end interaction diagrams are particularly convenient. In Figure XIII the interaction factor () for the combination X₁, X₂ as in point D will be equal to the ratio AD/BD in the case of synergism and to the ratio CD/BD in the case of antagonism. When iso- or hetero-addition give different outcomes, they should be used instead of the segment BD, and the upper and lower values of will be obtained, _u and ₁, respectively.

 

C. PROBABILISTIC ASSESSMENT OF THE INTERACTION

 

50. As any biological end-point which is expressed at a sufficiently high level of complexity may be viewed as the final outcome of a long chain of intercorrelated events, it appears quite natural and wholly justified to express any end-effect by the probability, P, that it may occur. The dependence of this probability on the absorbed radiation dose may sometimes be one of direct proportionality but in most instances, and particularly for the most complex effects, the relationship may be more complex. Essentially the same can be said about the effects of other physical and chemical agents.
51. In probabilistic terms, if two agents act simultaneously on a biological system, one possible assumption is that the two agents act independently, which allows comparison of the results of the joint actions with the presumed outcome of the two agents acting independently. In some respects, this notion is similar to that of heteroaddition, with the difference that the probabilistic approach is conceptually much broader. The discussion to follow will again be limited to the simple case of two agents acting simultaneously but may in principle be extended to any number of agents.
52. The Committee agrees that the most important effects of radiation in man are carcinogenesis and mutagenesis, effects that are described by the ICRP [I1] as stochastic in the sense that their probability of occurrence increases linearly with dose and without threshold; their severity is independent of dose; and no causal relationship with radiation exposure can empirically be established for any given case. It is well known that these effects do occur even in the absence of artificial irradiation with a frequency which is much higher than would be expected if they were induced only by natural background radiation. For the purpose of the present analysis, they may thus be viewed as stochastic derangements of physiological processes to which a probability of occurrence P0 could be attributed.
53. If one assumes that the exposure X₁ to an agent causes a probability P₁ of a given effect, t, the overall probability that this same effect can be observed, taking into account the spontaneous level P0 and assuming that P0 and P₁ are independent is:
    
                                                     P_t1 = P₀ + P₁ — P₀ P₁                                               (13)
    
    Since in biological experiments control and test groups are run concurrently, the following formulae may also be convenient:
    
                                                     P₀₁ = P_t1 — P₀                                                          (14)
    
                                                     R₁ = P_t1/P₀                                                                (15)
    
    

    They show the absolute and relative increase of the probability to observe the given end-effect following exposure X₁. A similar set of equations can also be written for a hypothetical second agent:
    
                                                     P_t2 = P₀ + P₂ — P₀ P₂                                                (16)
    
                                                     P_o2 = P_t2 — P₀                                                           (17)
    
                                                     R₂ = P_t2/P₀ (18)Pet                                                     (18) 
    
    

     

54. When the action of the two agents is combined, the expected probability of observing the overall effect P_et may again be calculated on the hypothesis of independent action:
    
                                                     P_et = P₀ + P₁ + P₂ — P₀P₁ — P₀P₂ — P₁P₂ — P₀P₁P₂    (19)
    
    The absolute and relative increases in probability of observing the effect as a result of the joint action of the two agents will accordingly be:
    
                                                     DP_exp = P_et — P₀                                                              (20)
    
    and
    
                                                     R_exp = P_et/P₀                                                                     (21)
    
    

55. When an experiment is performed on the combined action of two agents, the observed total probability of effect, Po_t, will in general be different from the expected probability Pe_t. One of the possible definitions of the interaction factor w might be simply the ratio between the actual and the expected probabilities, Po_t/Pe_t. It is easy to see, however, that if this definition is adopted the value of the ratio will depend critically on the absolute value of P0. When the effect under study has a high spontaneous level of occurrence, the interaction factor may be about 1 despite the observed absolute deviation between the experimental values and the expected value based on the hypothesis of independence. On this ground, another definition of the interaction factor, , is preferred, as follows
    
                                                     = DP_obs/DP_exp                                                           (22)
    
    where
    
                                                     DP_obs = P_ot — P₀                                                           (23)
    
    The two definitions of the interaction factor will naturally coincide if P₀ = 0. The probabilistic definition of in equation (22) coincides in essence with the definition of interaction factor in equation (5).

56. The denominator of equation (22) is calculated on the basis of the independence of action of the two agents (equation (19)). Equation (20) may be rewritten by using equations (14) and (17):
    
                                                               
    
    Accordingly, the equation for the interaction factor will assume the following form
    
                                                               = (P_ot — P₀) / (P_o1 +P_o2)                             (25)
    
    If Po₁ and P_o2 are small, the above equation reduces to 
    
                                                               = (^Pot — P₀) / (P_o1 + P_o2)                            (26)
    
    or, changing the probabilities to the corresponding ratios for P0 ' 0, according to equations (15) and (18)
    
                                                               = (R_obs — 1) / (R₁ + R₂ — 2)                      (27)
    
    where
    
                                                               R_obs = P_ot^/P₀                                                     (28) 

57. To give an example of such a type of treatment, the experiment on diploid yeast irradiated with alpha and gamma radiation [M3] and analysed by the method of the isobolic diagram (see section I.A.), may now be recalculated to obtain the interaction factor. For a dose of 9 Gy of mixed irradiation (25% alpha and 75% gamma) the level of reversion will be 180 per 10⁶ survivors, corresponding to a Pot = 18 10-⁵. The spontaneous level of reversion Po = 2 10-⁵. If one knows the alpha and gamma doses and the slopes of their regression lines, one may calculate Po_l and P_o2 to be 5.7 10-⁵ and 7.4 10-⁵, respectively. The last term in equation (25) is negligible and one may use equation (26)
    
                                                             
    
    
    The fact that is greater than unity suggests a synergistic interaction.
58. The question arises of establishing errors and limits of confidence for the interaction factor. The theory of error transfer may be applied to this end. The value of w as defined in equation (22) may be considered as the ratio of two stochastic quantities DP_obs and DP_exp. The mean value of this ratio is
    
                                                                                                                 (30)
    
    
    The error matrix for A P_obs and A P_exp will be
    
                                                             
    
    
    where S is a symbol representing the mean quadratic error and q₁₂ is the correlation coefficient. The mean quadratic error of will be in this case
    
                                                              

If the distribution of and DP_exp is normal and DP_exp/S(DP_exp) > 5, then the distribution of should also be approximately Gaussian.

59. The same problem of assessing an error to was considered by Rothman [Rl] for the case of epidemiological investigations and he used the same definition of the interaction factor. If a log-Gaussian sampling distribution is assumed, an estimator of (t) referring to a large sample interval may be written as
    
                                                             
    
    
    where k_a is the abscissa value for a given level of significance a. The methods for the calculation of S(1n ) for cohort studies and case-control studies have also been given in [Rl].
60. To illustrate further, the confidence limits for the experiment on yeast considered above [M3] in paragraphs 31 and 57 may now be calculated. In the example = 16 10^—5 and = 13 10^—1; s() = 0.8 10^—5; S() = 0.9 10^—5. In case of a synergistic interaction there cannot be a negative correlation between and ; equation (32) may therefore be used without the third term within brackets as an upper estimate of S²(). Fitting the above values to the equation, S²() = 0.01 and S() = ± 0.1. The estimated value is therefore = 1.22 ± 0.10 and, for 95% confidence limits, ₁ = 1.02 and _u = 1.42.
61. For complex biological systems a possible situation of isoaddition should be kept in mind. In probabilistic terms this means that if the action of agent 1 takes the system to probability level P*o₁, then P*o₂ will depend not only on the level of exposure X₂ but also on the value of P*o₁. Therefore P*o₂ becomes a conditional probability P*o₂(X2/P*o₁). The same is true for a reversed order of application of the agents. In this situation different conditional probabilities depending on the sequence of the agents and on the levels of exposure should be used in equation (24). This will give finally upper and lower limits for DP_exp. Corresponding upper and lower limits for the interaction factor w may be calculated as
    
                                                               _u = DP_obs/DP_exp                                     (34)
    
    These limits will be further extended by the presence of experimental errors.

 

D. THEORY AND PRACTICE

62. Before proceeding further to the analysis of some experimental and epidemiological data, it is necessary to discuss briefly the applicability of the concepts reviewed in the preceding section to situations involving complex biological effects. In doing so, it will immediately be realized that even problems which may appear of minor and mostly speculative importance in the analysis of the action of a single agent, are likely to become very difficult to disentangle when various agents are combined, giving rise to much uncertainty in the assessment of the type of interaction that might apply.
63. The definition of an effect is hardly ever a problem in radiation biology. The conditions under which the effect is manifested, its degree of expression or its probability of occurrence may usually be described with sufficient precision. What may be less easy to define is the dose-effect relationship at all levels of exposure, particularly at the low ones. When two agents are combined, depending on the form of the respective exposure-response relationships, more or less effect might be obtained at a given exposure regime than might be predicted on fragmentary knowledge of the relevant relationships. This points to the need to establish with sufficient precision through appropriate controls not only the response to be expected at the exposure levels of interest for the particular experiment, but to obtain a full dose-response curve for both agents under study. The ultimate aim is to establish experimentally the surface of response corresponding to the full range of both agents. However, if the number of experimental points to establish a given exposure response curve with one agent is N, to establish the whole surface of response with the same number of experimental points in a given sequence of administration requires N² experimental points. Reversing the order of administration will in turn double the number of experimental points to 2N². Such an increase in the size of the experiments is often not feasible and complete series of the type envisaged are almost never reported or conducted.
64. The definition of the level at which a supposed synergistic or inhibiting action may take place is extremely important in the analysis of such actions. Here the need for operational definitions of practical significance and the need of resolving mechanisms in biological experiments may often be at variance or even incompatible. If, for example, one takes a very complex biological short-term end-point, such as the death of an animal (but the loss of reproductive integrity of a cell may be sufficiently complex, depending on the level at which the mechanisms of action may be resolved), exposure to any toxic agent in sufficient amounts could produce such an effect. This of course will happen at times and with mechanisms differing from one agent to another and mostly specific to each agent. The combined application of two agents may in principle produce apparently antagonistic or synergistic effects when some of the pathways of action of the two agents happen to interfere with each other. But at this level of complexity, even though the end-point might be of practical significance, the real existence of combined actions may be difficult to assess. Only when the mechanisms of action of the two agents are reasonably well defined will there be any merit in making use of the concepts of synergism or antagonism, in order to avoid misuse of the terms. Within this framework it may also be discussed how the presence of one may enhance the detectability of another interacting agent, when both produce the same effect.
65. Confusion of iso- with hetero-addition could result in the false identification of synergism. For example, one could visualize two agents, both toxic to the bone marrow and both inducing leukopaenia with a very curvilinear relationship to exposure, as is usually the case. It is easy to imagine that the action of the combined treatments might produce more effect than expected by the same doses of the two agents separately, simply because of the curvilinearity of the relationships and of the isoadditive character of the combined effect. It is also easy to understand that death of the animals might ensue at levels of the combined agents which are much below those of the two agents acting separately. If leukopaenia and death were the end-points of reference, in the absence of any other information one might be tempted to think of a synergistic action. Yet, to call such an effect synergistic would be unjustified because isoaddition would be operating in this case. Clearly, without knowledge of the whole range of responses, it would be impossible to clarify the issue. It should be realized that too often the cases of synergism claimed in the literature have been insufficiently analysed in this respect and there is ground to doubt that they might stand up to more refined investigations.
66. As to long-term effects, it is usually thought that tumour induction is a sufficiently well-defined phenomenon to be taken as an end-point, as though all tumours have the same aetiology and pathogenesis and there are not great variations in the incidence of various tumours between species, strains and experiments. This assumption is imprecise when different doses of the same agent are administered, because expressing the response as overall tumour induction may mask important effects on some tumour classes. The assumption is however particularly dangerous in studies of combined actions because under these conditions changes in the tumour spectrum would certainly be expected. It is essential therefore that the end-point of the studies be specific and extremely well defined. The same reasoning applies to the genetic and developmental effects.
67. Changes in the state of the biological system may be brought about by sequential treatment. For example, a large body of evidence on mammalian cells indicates that dose fractionation in radiobiology is a difficult subject to investigate. Usually the first dose produces partial synchrony of the irradiated population, so that the response of the surviving cells to the second dose fraction is altered with respect to that of a non-exposed undisturbed population. It would be very easy but totally unjustified to think of antagonistic or synergistic effects in the absence of information on the survival curve of the overall population and of its constituent sub-populations and in the absence of data on the amount and time sequence of synchrony induced by the first treatment. There is every reason to believe that such cases may occur also in respect to other chemical treatments and it should in fact be pointed out that treatments with chemicals (BUdR, hydroxyurea, for example) are often used to obtain experimentally synchronized cell populations. The amount of information available in respect to such effects by the various agents discussed in the following parts of this Annex is lacking or extremely limited. Efforts to clarify the situations occurring in practice through experimental analysis might help to avoid misconceptions.
68. Another point calling for great caution concerns the time parameters of the combined action. Two types of treatment may be visualized, contemporaneous and sequential. Partial overlapping and fractionation of the exposure to each of the agents could increase the complexity of the temporal patterns of exposure. Contemporaneity of the treatment time does not necessarily imply a simultaneous action at the level of the target structures. For example, in the case of chemical or pharmacological substances, variable time for metabolic processing of the agents might be required and it would temporally displace the action on the biological structures of interest. If hetero-addition is assumed to operate, administration of one agent before another, or vice versa, should not in principle lead to a change in the end-result. But, on the other hand, if reversing the order of administration does produce a change (qualitative or quantitative) of the response, the conclusion should not necessarily be drawn that some interaction differing from additivity applies. This all points to the relativity of the definitions and to the difficulties of translating into sensible biological terms the precise statements of the theory.
69. There are biological effects for which the timing and the sequence of the actions is all-important. According to one hypothesis, for example, tumour induction may be regarded as the result of two independent phenomena, initiation and promotion. Initiation is visualized as a fast irreversible process acting on normal cells and conferring upon them the character of neoplastic ones. It precedes promotion but without the latter could not result in a growing tumour. Promotion, which on the other hand is ineffective if not preceded by initiation, takes place during fairly long times and may be reversible. Many agents share the properties of initiators and promoters in different degrees at different doses. Thus, reversing the order or altering the time pattern of administration of two carcinogenic agents is bound to produce changes in the qualitative or quantitative expression of their final action. This should be kept in mind when designing experiments on combined action.
70. The issues discussed in the preceding paragraph are further complicated by the fact that the time for tumour appearance is important, as is the final tumour yield. The rate of appearance of tumours in time (once this rate is referred to a given tumour type and is corrected for competing risks) is an important parameter since, in principle, it is related to the promotive action of a treatment; while the final tumour incidence is related to the initiation action. When agents possessing both properties are administered in combined experiments the precise nature of the interaction and its influence on the combined end-point would not normally be resolved without detailed information of the mechanisms involved.
71. The decisive importance of the temporal pattern of exposure to ionizing radiation vis-à-vis practically all biological end-points is documented for a variety of biological effects in the specialized sections of the previous report (see Annexes H, I, J of [U1]) and in Annexes I, J and K of the present report. In general, fractionation or protraction of the exposure lead to a decrease of the final effect, although in some cases deviations from this general pattern are reported [H19]. It is not unreasonable to expect that changes in the yield of effect may also occur by altering the pattern of exposure to other agents interplaying with radiation, so that the final effect of the combined treatment cannot be predicted, particularly in the region of the low doses which are of major concern in the present context. Precise information about the temporal distribution of the exposure is therefore required in evaluating the combined effects.
72. In conclusion, the notions of synergism, additivity and antagonism which may be defined in theory and evaluated by appropriate statistical analyses, are seen to lose some of their clarity when confronted with the complexity of biological organization and the variability of experimental conditions. They may, on the other hand, acquire important practical connotations. Normally the assessment of combined actions requires clear understanding of the nature of the biological effect under study; precise knowledge of the pattern of its manifestation in time for the combining agents; reasonable definition of the exposure-effect relation-ships for each of the interacting agents, particularly when effects must be analysed over a range of exposures; control experiments to check for the applicability of the effect to different conditions of exposure. Without the detailed information described above, such notions will probably remain confined to the realm of theory and the subject of disbelief or overestimation, as the case might be. Only studies of mechanisms might eventually solve these uncertainties.

[ Index | Next page | References ]