|Splitting more than hairs|
Linear no-threshold (LNT) is a hypothesized model of cancer induction in response to ionizing radiation. The model says that additional cancer risk is linear with respect to the absorbed dose, and becomes zero only at zero dose. This model is used as the basis of most nuclear-related legislation around the world, and in some chemical risk assessment.
LNT estimates that the risk of premature death from radiation-induced cancer is around 5% per sievert or 0.5% per 100 mSv of exposure. For reference, the average annual human dose from background radiation is about 3.1 mSv.
The effects of low dose radiation in humans are very difficult to establish, because the required sample sizes are very large and background cancer incidence is very high. However, the effect of high doses delivered in a short time is well known, mainly from the studies on Japanese atomic bomb survivors. The high dose data clearly indicates an increased risk of cancer for acute doses larger than 100 mSv, and is linear with respect to the dose. It was later augmented by data from studies of people occupationally exposed to radiation and from animal research.
The LNT model extrapolates the risk from high doses linearly to low doses, assuming that cancer risk is zero at zero dose. The justification for this is as follows:
- In vivo, there is a linear relationship between the radiation dose and double-strand breaks in the DNA, for doses from 1 mGy to 100 Gy.
- Each double-strand break is hypothesized to have the same probability of making the containing cell cancerous.
- Each cancerous cell is hypothesized to have the same probability of developing into a disease.
The main conclusion of this model is that no level of radiation is completely safe, and consequently radiation exposure must be reduced until it is as low as reasonably achievable (ALARA). A useful feature of the model is that the effect of radiation on a population can be estimated by summing together the individual doses into a collective dose. This leads to measuring total occupational exposure of a group of people to radiation in units such as man-sieverts. LNT states that a collective dose of 20 man-sieverts will cause one radiation-related cancer death, regardless of the size of the exposed group and the distribution of dose amongst its members. For example, it predicts that the 1×1011 bananas eaten each year, each of which gives a banana-equivalent dose of 1×10-7 Sv, must be causing about 500 cancer deaths annually.
Possible contradicting evidence
Since the LNT hypothesis was formulated, lots of new research has been conducted that doesn't conform to LNT's postulated dose response. One example is a study of lung cancer mortality in the United States by Bernard Cohen, which demonstrates that the human response to low level radon exposure is definitely not linear, though it cannot rigorously say what the actual response is. Informally, it appears to support radiation hormesis. A recent research finding indicates that low level radiation damage is repaired more efficiently, which would reduce the effectiveness of low dose radiation relative to LNT. Another study found that mice irradiated with 400 times background radiation over 5 weeks did not show any DNA damage, indicating that low dose rate radiation is less harmful, even if the total dose is the same.
Recently, it was discovered that Herman Muller, originator of the LNT hypothesis, had access to evidence that contradicted it, but nonetheless endorsed it. Apparently it was a matter of political expediency in an effort to ban above-ground atomic testing.
Some organizations, such as the Health Physics Society and the French Academy of Sciences reject using the linear no-threshold model to quantitatively predict the health effects of low level radiation exposure. Health Physics Society rejects the LNT-based concept of collective dose and says that it is wrong to say e.g. "if 2000 people get a CT scan (10 mSv), then one of them will die from cancer", while the French Academy of Sciences endorses the existence of radiation hormesis.
- Linear no-threshold hypothesis, Nuclear Regulatory Commission
- The Linear No-Threshold Debate in Radiation Protection, Kenneth L. Mossman
- USEPA: Dose-Response Assessment
- Evolution of ICRP Recommendations 1977, 1990 and 2007, p. 15
- HPS mentions the risk of dying is 5 in 100 000 per mSv. This gives 5% per Sv.
- Radiation fact sheet, Nuclear Regulatory Commission
- Bernard Cohen's web page
- E. J. Calabrese, Muller’s Nobel lecture on dose–response for ionizing radiation: ideology or science? Archives of Toxicology 2011, 85(12), 1495-1498. Full text PDF
- Health Physics Society: Radiation Risk in Perspective