Dear Arjun,
What is LNTH and will it make me better?
-- Baffled in Buffalo

Dear Baffled,

In olden times, LNTH was an acronym for Lavish Neighborhood of Troy in Hellas, which is the Greek word for Greece. The famous Helen of Troy lived there, but she ran off to Paris and Troy has never been the same since.

In the nuclear establishment, LNTH stands for Linear No-Threshold Hypothesis. It is a hypothesis that is used in regulatory practice to assess the cancer risk of low-level radiation. Low-level radiation is defined as a level of radiation dose that does not produce short-term observable effects like skin rash, vomiting, or high white blood cell count. Such observable (or somatic) effects are produced when a substantial radiation dose is delivered in a short time. Most somatic effects occur at doses of 100 rem or more, though white blood cell count changes occur at far lower doses. The same dose delivered over a period of weeks or months would not produce readily observable effects, except at the cellular level. Yet it could increase the risk of diseases (stochastic effects), of which cancer is the most studied.¹

Survivors of the Hiroshima and Nagasaki nuclear bombings have been intensively studied to estimate cancer risk. This has been a huge effort - more than 75,000 people have been studied for over 50 years - which is continuing. The estimates of cancer risk used in regulatory practice are largely based on the study of these survivors. However, since the survivors received rather large doses, and since their radiation dose was received over a very short period, extrapolating the risks to low dose levels delivered over long periods of time has proved controversial and difficult. Moreover, some researchers, notably the British physician, Alice Stewart, and her colleagues, have pointed out that the long-term survivors were probably among the healthier people to start with and this complicates extrapolation of cancer risk to the general population from the survivor group.

There are other sets of exposed populations. First, everyone is exposed to natural background radiation. There are also varying levels of exposure to indoor radon, which depends on house construction and on the region in which homes are located. The difficulty is that everyone is also exposed to many other risk factors, including natural and man-made environmental risks, diet, and heritable factors. Since there is a substantial rate of cancer due to all these other factors, it is very difficult to extricate the risks explicitly attributable to exposure to low-levels of man-made radiation, such as nuclear bomb fallout or radiation exposure in the workplace.

In this discussion we define cancer risk (R) as the expected value of the number of cases of cancer for a given radiation dose (D). Note that the risk of cancer incidence is about 50% greater than the risk of a fatal cancer. The various hypotheses discussed here do not specify a level of risk; they only deal with the shape of the curve that describes the risk in relation to dose.² (See the equations in the footnote.) There are other factors involved in risk determination including age and sex of the exposed person. Risk also varies by type of cancer. Specifically, risk factors for leukemia are calculated separately from risks of solid tumors, such as lung cancer and breast cancer.

The LNT hypothesis has been the commonly (though not universally) accepted way of extrapolating the risk of exposures at relatively high-levels to that at lower levels. The hypothesis states that a given increment of exposure to radiation, no matter how small, will produce the same increment of cancer risk. So if a person has a certain risk of getting cancer at one rem of exposure, his cancer risk would be doubled for an exposure of two rem, and halved at 0.5 rem. Further, if ten people collectively got one rem, their collective risk would be the same as that of one person being exposed to one rem.

Collective population exposure is expressed as person-rem, which is the sum of all individual exposures in a population. From an estimate of collective dose, one can then apply a constant risk factor to get a statistical estimate of the number of additional cancers that would result from that exposure. In US regulatory practice it is common to assume that the risk of a fatal cancer in a population equals about one excess fatal cancer for every 2,500 person-rem of exposure. Figure 1 shows the LNT hypothesis.

There are other hypotheses about the shape of the dose-response curve. The most common alternative no-threshold hypothesis is the "linear-quadratic" hypothesis. According to this, there is a risk term that is directly proportional to the dose (the linear term) and another proportional to the square of the dose (the quadratic term). Figure 2 illustrates a quadratic dependence of risk on dose (linear term equal to zero).

There are those who believe that there must be a threshold below which there is no increase in cancer risk. They argue that some toxic materials exhibit such thresholds and that radiation has one too. Such thresholds may derive, for instance, from the ability of the body to repair damage caused by lower doses of radiation. Figure 3 shows a threshold hypothesis, with a linear risk response for doses higher than a threshold of T rem. However, it has been pointed out that since human beings are already exposed to natural radiation as well as other natural and artificial exposures that stress the body's repair system, the linear no-threshold hypothesis may, in any case, apply to radiation doses imposed by human activities because they are increments to other exposures. Hence, for the purposes of estimating the risks from human activities, the LNT hypothesis could still be valid and is a sound basis for public health protection.

There is also some evidence from recent experiments that low doses may produce a higher level of risk per unit of dose.³ This is known as the supra-linear hypothesis, and is shown below in Figure 4 . Finally, there is the "hormesis" hypothesis, according to which a small amount of radiation could produce some beneficial health effects, by stimulating the immune system for instance. The main evidence put forward for this has been from experiments on mice. According to a summary of the evidence for the hormesis effect, compiled by Charles Waldren, a high dose of radiation produced fewer mutations in some circumstances if preceded by a dose in the 1 to 20 rem range. This supposed protective effect does not appear at lower or higher doses, however, and lasts only for about a day, after which it disappears. Such a hormesis effect, even if it exists in humans, has no public health significance, especially in view of the evidence for other long term risks produced by doses of a few rem.⁴

The vast majority of work on radiation risk has been focused on cancer. There are a number of other potential risks (see letter to the BEIR VII committee). It is possible that non-cancer risks could, at least for some people and in some circumstances, be more severe than cancer risks.

Many of those who have put forth arguments for the threshold and hormesis hypotheses have also been arguing for a relaxation of current radiation protection regulations.⁵ This would be highly inappropriate for several reasons. First, there is considerable uncertainty about the health effects of low-level radiation. It is sound public health practice in such circumstances for regulations to err on the side of being more stringent. Second, the risk of radiation has, over the decades, been consistently revised upward. Even though that might not continue indefinitely, it is reason enough not to relax standards or to discard the LNT hypothesis. Third, there is evidence that the response to radiation varies widely among individuals. Standards should be set to protect the more vulnerable populations. Fourth, even if there is a threshold, it is important to remember that regulations are about additions to radiation. The linear no-threshold hypothesis would still be appropriate to assess excess cancer risk - that is the risk imposed by incremental radiation doses. Fifth, there are many non-cancer effects and synergistic effects that are not yet well researched; some are not yet researched at all. Finally, some of the potentially affected groups are among the most vulnerable to the ill effects of exposure (see letter to the BEIR VII committee). Stringent regulations based on a linear no-threshold hypothesis provide a modicum of protection for non-cancer risks and to vulnerable groups, until such effects can be carefully researched.

There are therefore sound reasons to continue to use the linear no-threshold hypothesis for regulatory purposes. When the questions such as the ones we have raised are answered properly, there will time enough for discussion about revising standards.

Dose Rate Effectiveness Factor

There is some evidence, mainly from animal experiments, that low radiation doses delivered at low dose rates produce a lower risk than the same dose delivered all at once. This supposed lower effectiveness of low dose rates is express by a factor called the Dose Rate Effectiveness Factor, or DREF. The adjusted risk per unit of dose for low dose rates is obtained by dividing the unadjusted risk by the DREF. It is general regulatory practice to assume that the risk from low dose rates is lower than the unadjusted risk by about a factor of 2. Hence the US EPA applies a DREF of 2 to the unadjusted BEIR V cancer risk coefficient of 0.08 fatal cancers per person-sievert to get an adjusted risk of 0.04 fatal cancers per person-sievert. The latter figure is the risk factor used in current radiation protection regulation in the United States. (1 sievert = 100 rem)