The linear hypothesis: have we reached a turning point?

1 January 1998

The firmly established “linear no-threshold” (LNT) hypothesis for cancer induction from radiation exposure has been attacked by heretics who argue that in low doses, radiation may cause no harm and may even have a positive effect. At a recent conference in Seville, Spain, on the biological effects of low doses of radiation and their regulatory control, it appeared that the diverging opinions are more akin to religious than to scientific beliefs. The present confusion at least seems to recognise that there is a problem and may represent a turning point that could lead to a more rational system.

The adoption by the radioprotection authorities of the linear no-threshold (LNT) hypothesis for cancer induction from radiation exposure, with its assumption that even the lowest doses have the potential to cause harm, has come under increasing criticism in recent years. There is growing support for view that there is, if not a threshold, at least a dose below which radiation causes no apparent harm, or even that in low doses radiation may have a beneficial effect. The problem is that there is little direct evidence to support either case.

The conference, “Low doses of ionising radiation: biological effects and regulatory control”, held in Seville, Spain, 17-21 November and jointly organised by the IAEA, together with the World Health Organisation and with the co-operation of UNSCEAR, was the latest of the attempts to present a sound biological basis for radiation protection. To this end the organisation of the meeting was firmly weighted in favour of the official belief in the LNT hypothesis. The tone for each of the ten conference sessions was set by official “keynote” speakers presenting papers of half an hour or more.

The heretics were effectively silenced; they were in some limited cases allowed to express their views in three minute contributions, in questions from the floor, or have them buried among some 200 other contributed papers of varying quality and interest which were, however, usefully published in a 4 cm thick book presented to all participants. Complaints that reputable bodies who had previously criticised the LNT theory, such as the French Academy of Science and the American Health Physics Society, had been excluded were rejected on the grounds that official representation at the meeting was determined by the national authorities – but these no doubt would be guided by their own radiation protection organisations.

Although under these one-sided circumstances there could be little real debate, the meeting provided an opportunity to survey some of the new developments in radiation biology which are now emerging. But even as the techniques of DNA, chromosomal and genetic analysis reveal more of the mechanisms and details of chromosome mutations, any direct relation to cancer, when the growth of a tumour is a process that may involve some six, or up to twelve, further genetic mutations at different sites over some time, remains tantalisingly remote. There is also some evidence that the response may in addition depend on an individual’s genetic make up. As professor Hall of the Center for Radiological Research, Columbia University (USA) said in his keynote paper: “that so many mutations could be caused by a single modest dose of radiation is, to say the least improbable. Yet there is hard evidence from the Japanese survivors of the A-bomb that a prompt exposure to radiation can induce a whole spectrum of malignancies including cancers of the digestive tract.”


Epidemiological studies of the Japanese survivors, one of the largest, and most studied, groups of those who have been exposed to radiation has hitherto provided some of the strongest support for the LNT theory. Yet here also there are uncertainties. To what extent can the extrapolation downwards from results seen in those exposed to high radiation doses in a very short burst be relevant to exposure at low doses over extended periods of time? A paper from D L Preston of the Radiation Effects Research Foundation, however, pointed out that some 80% of the Japanese survivors received doses that were less than 100 mSv. He then claimed that the latest Life Span Study data show “a significantly increasing trend in cancer risks over the range 0 to 50 mSv.” This conclusion however seems to be arrived at by adjusting the raw data with a complex procedure. There are also the uncertainties in estimating or attempting to reconstruct the doses actually received by those exposed to the bombs. A M Kellerer of the Munich University Radiation Biology Institute pointed to yet a further possible revision of the dosimetry on which these estimates are made following two previous ‘official’ changes. If accepted this latest change would suggest that the cancer risks may have been overestimated.

In her keynote paper on this topic, E Ron pointed to some of the difficulties in reaching reliable epidemiological conclusions: uncertainties in dose estimates particularly in looking back to past exposures for case-control studies; the need for large numbers over a long time for cohort studies; and the limited value of ecological studies making comparisons for geographical regions with different exposures. A heretical contribution by T D Luckey which met the requirement for large numbers by combining the results of nine studies of exposed nuclear workers and other populations was dismissed, presumably because it came up with the ‘wrong’ answer – that low level radiation had a beneficial effect on cancer induction.

The ecological studies, comparisons of populations living in areas of high and low background radiation, have also failed to provide any clear evidence. Yet the range of such exposures can be large. Dr Kesevan of the Bhabha Atomic Research Centre reported radiation levels in the Kerala area on the south-west coast of India where the average dose received is nearly four times the normal background radiation level with actual levels ranging from 1 to over 35 mGy a year. No discernible increase was found in the incidence of chromosomal aberrations of new-born babies whose ancestors have lived in the high background area for hundreds of years. It may be irrelevant but this report also noted that the state of Kerala, with the highest literacy in India, has the lowest infant mortality and the lowest family size.

Some hope was expressed that valuable epidemiological information might be obtained from the large populations exposed in the former Soviet Union from the Chernobyl accident. But with the number of papers from those countries reporting increasing incidence of suicide, accidental death, and other diseases leading to a rapid decline in life expectancy – attributed to increasing social and psychological stress – the possibility of identifying specific effects due to radiation must be remote.


A topic of particular interest is the recent work on adaptive responses – mechanisms whereby very low (conditioning) doses of radiation appear to reduce the effects of damage by subsequent high doses. Mounting evidence for this, worldwide, led UNSCEAR to accept in 1994 that there is no doubt about the phenomenon’s occurrence at the cellular level even though they were reluctant to accept that it had practical consequences in estimating damage to populations and, in particular, in estimating cancer risks.

Similar adaptive response effects have been found with chemical agents. Radiation mimetic chemicals, alkylating agents, cross-linking agents, as with ionising radiations could all lead to a decrease in the cell’s sensitivity to the same agent or any of the others. From this it is suggested that the adaptation induced was attributed to novel, efficient chromosome break repair mechanism that if present at the time of challenge by a high dose would lessen the damage after such exposure.

It has not yet been possible, because of the long latent period for cancer induction, to determine whether adaptation does indeed lower cancer rates in pre-exposed human populations or increase survival rates, but preliminary animal experiments suggest that this might be the case. Those who adhere to the LNT belief, however, dismiss this approach on the grounds that even a low conditioning dose of radiation would have a harmful effect and should therefore be avoided.

Given the scientific uncertainties it is not surprising that doubts are now emerging on the way in which the assumptions of radiation damage are applied to radiological protection. The problem arises from the implication of the LNT theory that any radiation exposure, no matter how low the dose, can cause some additional risk of harm. This has led to the imposition of dose limits which are below the much larger variations in natural background. This exclusion of the higher natural background exposures from regulatory control is reflected in an ambivalent public attitude; there is a clamour for the strict application of limits to any nuclear facilities – at someone else’s cost – but little willingness to spend one’s own money on measures to reduce, for instance, much larger doses from radon.


Some of these doubts emerged at the Seville meeting with three novel approaches being outlined. Geoff Webb of the IAEA suggested a compromise solution whereby the LNT relation could be retained but with less weight applied to the low dose region to take account of a possible non-linear or adaptive response. Such an approach had previously been indicated by Sir Richard Doll, the doyen of epidemiologists, who was quoted in a report to Seville on the earlier low dose meeting in Stratford-on-Avon, UK, last May, as saying, “I continue to postulate a linear relationship at low doses, though with a slope at doses less than (say) 10 mSv that is less steep than that observed at moderate doses.”

A more far-reaching proposal came from A Gonzalez who presented the initial proposals of an ICRP task force on chronic exposure. This drew on a distinction between ‘practices’ and ‘intervention’. Practices include the production, processing, handling, storage etc of radioactive substances; intervention applies to radiological emergencies and lasting exposure from past or old practices. While the aim of the regulators is to reduce exposures as far as possible, they should seek to control only the additional dose, not the total dose. In the case of practices, which presumably bring some benefit (otherwise they would not be introduced) the dose which is additional to background should be strictly controlled in accordance with the present ICRP recommendations with a dose limit set at 1 mSv/year, dose constraint from 0.3 mSv/year, optimisation from 0.1 mSv/year (top left diagram). (Essentially, constraint means that action is required, and optimisation means that one should think about action.)

For intervention it is a matter of dose reduction through remediation measures which should take account of normally higher background (ie existing) levels. Here much looser limits apply rising in three bands from a normal level of 1-10 mSv/year, to a level of concern between 10-100 mSv/year where some assessment would be required, with a danger level at 100-1000 mSv/year (top right diagram). The logic of this dual limit approach is that an intervention reduces an existing dose (bottom right diagram), whereas a practice inevitably increases a dose (bottom left diagram). For a reduction of an existing dose, the more relaxed limits of the right hand diagrams apply, whereas a practice which gives rise to an increase in dose is subject to the more stringent limits in the left hand diagrams.

There must however be doubts how two different sets of limits, differing by up to three orders of magnitude, can be maintained or explained to the public when the outcome depends on determining whether the exposure comes from a practice or an intervention – a distinction that may not always be obvious.

The third proposal came from Roger Clarke of the National Radiological Protection Board who put forward a range of ‘controllable’ doses based on risk estimates. Controllable doses would for instance exclude natural background. At a risk of 1 in 1000 per year, a tolerable level, the controllable dose should not exceed 30 mSv, but this is not a limit in the sense of being a boundary between safe and unsafe but a level above which action should be taken; above 3 mSv per year with a corresponding risk of 1 in 10 000 some remedial steps may be needed; 0.3 mSv per year with a risk of 1 in 100 000 is the maximum dose to an individual who receives no direct benefit, while 0.03 mSv when the risk is 1 in a million a year is regarded as trivial. The status of Clarke’s proposal was not clear. It was made in the final conference session when he presented his summary report as one of the session chairmen; there was then no opportunity for discussion or comment.


There was also criticism of the use of the concept of collective dose when it is used to multiply a minute individual dose by a very large number of individuals exposed to reach numbers that can then predict actual cancer cases – as had been widely done in the case of exposures in countries of western Europe to Chernobyl fallout. Collective dose, L-E Holm, the head of the Swedish Radiation Protection Institute, insisted should only be used as a decision making tool and not as a prediction of actual health effects. But here again those who are committed to the linear relationship down to zero dose could argue that any individual dose, no matter how small, has the potential to cause harm which could on a statistical basis lead to real numbers of cases.

While the meeting did little to resolve or even give an equal opportunity to debate the questions of radiobiology, it is encouraging to see that some movement is being made towards the adoption of a more rational approach to radiation protection.

Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.