int. j. radiat. biol 1998, vol. 74, no. 6, 799 ± 804
An illustrative comparison of the event-size distributions for c-raysand a-particles in the whole mammalian cell nucleusK. BAVERSTOCK†* and M. THORNE‡(R eceived 21 M ay 1998; accepted 17 J uly 1998)
Abstract. agents) by relating the probability of developing aPurpose: Recent laboratory studies of endpoints designated as due cancer to the exposure of the target tissue to radi-to radiation-induced genomic instability have cast doubt on the ation, expressed in terms of absorbed tissue dose.validity of the current theoretical framework. Under this frame- This applies for certain qualities of radiation, mainlywork extrapolations are made from directly determined risks of low LET, within a limited range of mainly ‘high’radiation-induced cancer to those circumstances for which nodirect information exists, namely at low doses and dose rates at doses and dose rates. Radiation is most commonlylow LET and at low dose exposures to high LET radiations. understood to lead to cancer through the accumula-Based upon an approach in which the ‘state’ of the genome, as tion of mutations to speciŽ c genes, many of whichexempliŽ ed by the pattern of gene expression, rather than the are supposed to be responsible for growth controlbase sequence of the genomic DNA, is taken to be the origin of and ‘genomic housekeeping’. The absorption of radi-genomic stability, it is hypothesized that the critical factordetermining the likelihood of destabilization by ionizing radiation ation leads to ionization, either directly in the DNAis the dose to the whole cell nucleus. of these genes, or resulting in the generation ofCalculations: The frequency distributions of event sizes from two reactive free radicals from other components of thequalities of radiation, low LET c-rays and 5MeV a-particles, nucleus, including water, which chemically attackare compared with 60Co c-rays being taken to be a low LET the DNA. The resultant chemical damage leads toreference radiation in determining the RBE of other radiationqualities. In the absence of measured event-size distributions for changes in the genomic sequence, either disabling,60Co c-rays in spheres of similar size to the human cell nucleus, activating or over-expressing speciŽ c genes. Thesethe 4.5mm sphere has been chosen as illustrative. Frequency changes are assumed to be inherited in a stabledistributions for 5MeV a-particles are derived, based on the manner in progeny of the irradiated cells, so creatingidealized situation of a parallel beam of constant LET, with all a clone of cells, which through further accretion ofparticles traversing the sphere.R esults: When compared for a dose of 1mGy the event-size (dose) mutations will, after a latent period of up to severaldistributions of the two qualities do not intersect. It is estimated tens of years, express the malignant phenotype.that only 0.4% of the energy from the a-particles falls in the The probability of cancer induction is often lin-range of event sizes that can be produced by 60Co c-rays. early related to absorbed tissue dose, or more-or-lessConclusions: Contrary to belief over the past 50 years, there is, in so, in an irradiated tissue or organ, and very rarelythis low dose range, no ‘continuum’ based upon quantities suchas LET or lineal energy that would provide a basis for extrapola- according to a power of dose greater than two. Thistion from measured RBE values. RBE is thus seen to be purely can be interpreted as indicating that the initialempirical. In addition, the potential to induce e å ects in bystander change, leading to malignancy, takes place in a singlecells is not considered when deriving weighting factors for a- cell from which the malignant clone is derived, i.e.particles of the type that contribute signiŽ cantly to public cancer is ‘monoclonal’ in origin. It is clear that cellsexposure to environmental radiation. receive signals from their neighbours that in uencetheir growth control, but if it were necessary to1. Introduction induce relevant changes in several cells at the sametime, the e å ects would be related to higher powersRadiation is a reproducible, if somewhat weak of dose.carcinogen, which through epidemiological studies The long latency for radiation-induced cancerson exposed populations has been relatively well and the fact that malignant cells exhibit degrees ofquantiŽ ed (compared with many other carcinogenic malignancy, both point to cancer being a progressiverather than an ‘on/o å ’ process. Conventional radio-*Author for correspondence: e-mail: 106305.2424@compuserve.com biology sees the latent period as one in which a†WHO REMPAN Project Oæ ce, WHO Collaborating ‘transformed’ or ‘initiated’ cell acquires the additionalCentre, Authority for Radiation and Nuclear Safety, PL/PO speciŽ c mutations that ultimately result in the malig-Box 14, FIN–00881, Helsinki, Finland. nant karyotype and phenotype. Under most models‡Hollybrook, 38 Pondtail Road, Horsham, West Sussex RH125HR, UK. these mutations are considered independent of one0955–3002/98 $12.00 © 1998 Taylor & Francis Ltd

800 K. Baverstock and M . T horneanother, i.e. the acquisition of one mutation neither without explanation of what this might involve at thedetailed molecular level. More recently, it has beenpredisposes nor protects against further mutations.Thus, the probability of full expression of malignancy shown that human epithelial cells, when irradiated,release into the culture medium factors that canis the product of the individual probabilities of allthe contributing mutations. a å ect the clonogenic survival of cells that have notbeen irradiated (Mothersill and Seymour 1997).It is within the framework of the understandingoutlined above that it is necessary to extrapolate Subsequently (Seymour and Mothersill 1997), it hasbeen shown that the medium from the irradiatedfrom the primary data upon which risk from exposureto radiation is estimated, to obtain the risks from low keratinocytes can induce delayed e å ects in the sur-vivors of the initial exposure to the medium. Thesedoses, low dose rates and radiation qualities otherthan low LET, such as a-particles from actinides. so called ‘bystander’ e å ects are presumably mediatedby some kind of signalling between cells and it isThe primary data are derived from the quantitativeepidemiology relating to the e å ects seen in the these signals and not the ionizing radiation thatinduces lethality and delayed e å ects similar to thosesurvivors of the atomic bombings in Japan, who wereexposed overwhelmingly to energetic c-rays with associated with genomic instability.The observed features of radiation-induced carci-doses up to several Gy in a fraction of a second.Some epidemiological studies, which do not provide nogenesis, i.e. monoclonal origin, long latency, stageddevelopment, etc., are equally, if not more consistentprimary data, can act to endorse the results of theextrapolation process – in particular the study of with a rather di å erent framework for cancer induc-tion by radiation, namely that radiation, either dir-workers in the nuclear industry (Cardis et al. 1995)where dose rates were several orders of magnitude ectly or indirectly through cell-to-cell signalling,induces instability in the irradiated genome and thatlower than for the survivors of the atomic bombingsin Japan, although accumulated doses were compar- malignancy is one of the possibly many fates of thedestabilized genome. This framework di å ers in aable. However, at doses of the order of a few mGydelivered over days to years and for other qualities number of important respects from the more conven-tional one outlined above, for example the changes,of radiation, such as a-particles and neutrons, atsimilar doses and dose rates, few if any conŽ rmatory including much of the molecular and chromosomaldamage, that occur are irrelevant to the endpoint.data sources exist though there are data for internallyincorporated a-emitters delivering very high radi- In particular, it is not a mechanism so much as aprocess in which the individual steps in a given cellation doses (UNSCEAR 1994).Increasingly, evidence that the framework outlined are merely points along one of many possible routesfrom normality to malignancy, bearing no particularabove might be  awed is accruing. Loeb (1991) notesthat unless the initial mutation in the sequence of relevance to the endpoint. Secondly, these steps arenot necessarily independent from one another. It ismutational events leading to malignancy increases theoriginating cell’s susceptibility to further mutations, not the purpose of this paper to explore this processin detail, but rather to explore the implications thatthe probability of acquiring the necessary mutationsfor the malignant phenotype would be so low as to such an alternative framework would have for oneimportant aspect of radiation protection: the choicemake the development of cancer an impossibility.Loeb therefore proposes that the Ž rst event endows a of radiation weighting factors to account for thedi å ering e å ectiveness in inducing cancer of variousso-called ‘mutator phenotype’ on the cell and thatsubsequent mutations can be acquired more easily. qualities of radiation.That mutations would be independent of one anotherseems unlikely in an entity such as the genome, in 2. A hypothesisany case, given the complex inter-dependenciesobserved in other aspects of cellular metabolism. For the purposes of this argument it will beassumed that radiation destabilizes the genome byFurthermore, the observation of radiation-inducedgenomic instability, in which cells that have survived causing it to mobilize defences against the moleculardamage caused by the ionizing energy deposited byirradiation and divided several times apparently pro-duce de novo damage associated with the exposure the passage of a particle or photon in the immediatevicinity of genome. For example, the delaying of(Morgan 1996), challenges the concept that mutationsinduced by radiation are only inherited by replication mitosis (Bernhard et al. 1995), the activation of repairprocesses (Le et al. 1998) and possibly apoptosis, haveof the mutational damage that was ‘Ž xed’ at the Ž rstmitosis after exposure. This phenomenon of late all been established to be typical responses of the cellto large amounts of molecular damage caused byexpression of radiation-associated e å ects is attribu-ted to radiation-induced instability of the genome, radiation. In these processes the expression of certain