Copyright R. Zamenhof, 2013.
Kerala beach, India, which has one of the highest terrestrial background dose rates in the world due to the abundance of thorium-containing monoxite sand. Dose rates are 7 mSv/year (700 mrem/year) compared to average terrestrial dose rates in the U.S. of 0.3 mSv/year (30 mrem/year). Despite the Kerala terrestrial dose rate being approximately 20x higher than in other areas of India, epidemiological studies have not detected elevated cancer rates in residents of Kerala compared to the rest of India.
It has been estimated that about 10% of genetic mutations that occurred during the evolution of human life have been due to the influence of radiation. We are all exposed to natural background radiation on a daily basis: cosmic rays from outer space that interact with our atmosphere and shower us with various secondary radiations; gamma-rays produced by radioisotopes naturally present in the earth; radioactive radon gas that oozes out of the ground and enters our lungs; and two or three radioisotopes that reside naturally in our bodies. Since evolution is driven by genetic mutations, natural background radiation would not appear to be that bad for the development of the human race. However, genetic mutations are a two-edged sword: they help drive evolution according to the processes of natural selection by enhancing the selective survival of “desirable” genes, but they can also cause illnesses such as cancer. This article will consider the latter of these effects of radiation, i.e., those that are potentially detrimental to human health even though they may sometimes be unavoidable.
Radiations we are exposed to
Let’s review in more detail the physical nature of the radiations we are exposed to. These consist of two major categories: electromagnetic radiations, and particulate radiations. Electromagnetic radiations include x-rays and gamma-rays, while particulate radiations include alpha-particles, protons, neutrons, and electrons.
In addition to natural background radiation, referred to earlier, we are also exposed to man-made diagnostic radiations and therapeutic radiations, used widely in the medical area.
Radiations used for diagnosis of disease
Diagnostic x-ray machines take two-dimensional x-ray images of your chest and of many other body parts, whereas highly complex and sophisticated x-ray machines such as CT scanners produce x-ray images of your body in the form of thin (1-4 mm) thick slices, eliminating the problems of tissue overlap that inhibit accurate diagnosis in the plain-and-simple two-dimensional type of x-ray imaging.
Radiations used for therapeutic treatment of disease
Linear accelerators, descendants of radiation-generating equipment used for many decades in physics research, produce very high-energy x-ray and electron beams that are used to treat cancer. Electron beams lack the high “aiming” accuracy possessed by x-rays and gamma-rays, but they weaken and disappear very rapidly at depths beyond the boundaries of relatively shallow tumors, thereby protecting radio-sensitive tissues or organs that may be located downstream of the tumor.
Gamma-rays, produced by radioisotopes encapsulated in rice-sized metallic “seeds” that are inserted into some types of tumor, are often used to treat cancers such as prostate or breast cancer from inside the body, where tumors may be surrounded by particularly radiation-sensitive tissues or organs. A machine called the Gamma-knife uses gamma rays to treat primary brain tumors from outside the body, as well as tumors that have spread (metastasized) to the brain from cancers at other anatomical sites.
The use of protons beams is a relatively new development in radiation therapy. There are currently about 35 sites in the U.S. that offer this form of radiation therapy, largely due to the extraordinary high cost of such facilities: typically $100m-$150m (although more recently developed “single-room” proton facilities are less costly). At the present time, proton beams are most commonly used for treating prostate cancer in adults and brain and spinal cord tumors in children, although other types of cancer are treated as well. Some cancers are difficult to treat with x-rays or gamma-rays because tumors may be surrounded by especially radiosensitive tissues or organs, limiting the amount of radiation that can be delivered to the tumors themselves. To a large extent, proton beams sidestep this problem. Unlike x-rays and gamma-rays, which penetrate the entire body and in doing so deliver potentially damaging radiation to healthy tissues downstream of the tumor location, when proton beams reach their target they abruptly stop, completely avoiding the downstream radiation exposure problem. Electron beams also stop after reaching a specific depth—although not nearly as abruptly as proton beams—but they also spread out laterally, whereas proton beams have exquisite aiming accuracy, both in depth and in lack of lateral spread, so enormously reducing radiation exposure of healthy tissues.
About two-thirds of the naturally occurring background radiation dose we are exposed to on a daily basis consists of alpha particles. Naturally occurring alpha particles are produced by radon gas (radon gas is the decay product or “daughter” of radium which resides in the superficial layers of our planet) that we absorb into our lungs with every breath we take. Radon gas, which is highly radioactive, oozes out of the ground and building materials mixing with surrounding air, so breathing it into our lungs is unavoidable. Alpha particles have a very short range in tissue—just a few cell diameters—so when inside the lungs they dump all their energy in the very thin layer of epithelial cells lining the lungs. The large mass, high electrical charge, exceptional ability to produce un-repairable biological damage, and the short-range of alpha particles produces more “bang for the buck” in terms of radiation damage than any other type of radiation.
How Can Radiation Cause Cancer?
Now that we’ve reviewed the nature of the radiations that we are exposed to, let’s think about how these radiations could cause cancer. Cancer, as far as we know, is the result of genetic mis-programming caused by mutations in our DNA. Such harmful mutations can occur naturally due to random processes, or they can be caused by external environmental factors such as chemicals or radiation.
Epidemiological Evidence for Radiation Risk to Humans
The evidence that we have on the relationship between human radiation exposure and cancer comes from man-made radiation sources to which humans have been inadvertently exposed. These include the Hiroshima and Nagasaki atomic bombs, irradiation of the spine that many decades ago was a standard treatment for a congenital disease called ankylosing spondylitis, and irradiation of the female breasts in tuberculosis sanatoria, mainly in Massachusetts and Canada, where a standard therapeutic approach was to deflate and re-inflate the lungs under x-ray fluoroscopic guidance. The theory at the time was that this maneuver would deprive the tuberculosis-causing microorganism of oxygen; but today we know that the responsible microorganism is anaerobic, i.e., does not require oxygen to survive and multiply. The x-ray fluoroscopic equipment in those early days produced hundreds of times more radiation dose to patients than modern fluoroscopic equipment, and because these patients received lung deflation and re-inflation under fluoroscopic guidance on a monthly basis, resulted in massive amounts of radiation dose being delivered to the breasts which, in turn, resulted in a measurably increased rate of breast cancer.
Derivation of Radiation Risk Models from Historical Radiation Effects Data
Statisticians working for U.S. and European organizations such as the NCRP, ICRP, ICRU, and the ABCC got hold of the above-mentioned data and drew a straight line, originating at zero radiation (where, presumably, zero additional cancers were caused) and passing through the average of the very scattered data points relating radiation dose to the incidence of cancer. This straight line is referred to as the “linear no-threshold radiation effects model”, or the “LNT model”. Since there are no actual data points at the relatively low radiation doses that are characteristic of natural background radiation and diagnostic x-ray doses, the LNT model is only a theoretical predictor of what additional cancer cases might be expected at these lower radiation levels. The gradient of this LNT line provides the only relationship we have linking radiation dose to cancer incidence and cancer death. For example, the number on the right in the table below shows the slope of the LNT line relating additional (excess) cancer deaths to radiation dose, in units of lifetime excess cancer deaths per 10,000 members of the general public exposed one time to 1 rem (1 Sievert in modern units) of radiation; we will not differentiate between the units of rem and rad or Gray and Sievert for the purposes of this discussion.
*Exposure received only after age 18 years. Data are weighted averages; i.e., the older you are at the time of the single exposure, the lower your risk, since you have less years left for the effect to express itself. The actual risk values change a little from report to report but are basically as shown.
This means that if 10,000 members of the general public were exposed to 1 rem of radiation dose, then within the remaining lifetimes of these individuals, 5 would contract fatal cancers statistically caused by that 1 rem of radiation dose. Now one can argue that those 5 cases of fatal cancer would equally likely to have been caused by natural background radiation or by non-radiation carcinogens such as chemicals. This is a totally logical assumption, and is the reason why it is very difficult to establish in tort law that a certain radiation dose caused a specific individual to die of cancer.
However, one can advance an epidemiological argument that each of the people exposed to 1 rem of radiation would have a probability of contracting a fatal cancer from that 1 rem of radiation dose that is (1 X 5 / 10,000) x 100, or 0.05%. Now the natural fatal cancer rate among the human population in the U.S. and Europe is approximately 20%, which means that 1 rem of additional radiation dose raises that probability to 20 + 0.05, or 20.05%. Expressed from that perspective, 1 rem hardly seems like a dose to be enormously concerned about.
Computation of Risk Estimates Using the Standard LNT Radiation Risk Model
The final piece of the puzzle we need to consider is the actual radiation doses produced by various sources of radiation that we can plug into the LNT equation and assess the corresponding risk.
Consider two illustrative cases
1) The doses from diagnostic radiology procedures that use x-rays range very roughly from 1 milli-rem (for a standard chest x-ray) to 1,000 milli-rem (for a typical CT scan); or, in modern units, 10 micro-Sievert to 10 milli-Sievert. Using the LNT gradient parameter “5.0” from the table above, for 10,000 exposed people, this dose range corresponds to between (5 x 0.001) and (5 x 1), or 0.005 to 5 additional fatal cancers per each x-ray procedure per 10,000 exposed people. Expressed as a percentage probability for a single individual, this corresponds to [0.005 x 100] / [10,000] to [5 x 100] / [10,000] = 0.00005% to 0.05%.
2) If the radiation dose received by 10,000 members of the general public were only due to one year’s worth of natural background radiation (which averages around the U.S. to roughly 300 mrem/year; or 3 milli-Sieverts/year), the number of additional fatal cancers caused in that population of 10,000 people would be 5 x 0.300, or 1.5 additional cancers/year for each year that each person was exposed to background radiation, or equivalently [5 X 0.3 X 100 / 10,000] = 0.015%. So since we are continuously exposed to background radiation, a 50 year-old individual exposed to natural background radiation from birth would have a [50 x 0.015%] = 0.75% likelihood of contracting a fatal cancer from natural background radiation.
One can conclude, therefore, that for an individual person the risk of fatal cancer induction due to radiation, either from natural background or from diagnostic x-ray procedures, is still very small compared to the baseline fatal cancer incidence rate of 20%.
Some Limitations of the LNT Cancer Risk Model
Despite what has been said, a number of mitigating factors need to be stressed.
1) The calculations presented here are based on the LNT cancer risk model. Even using the most extensive human radiation effects data we have at the present time, there are very large statistical uncertainties associated with such calculations—often larger than +100%; i.e., the probability of 0.05% cancer incidence due to a single CT scan, when expressed as a statistical range, would be zero - 0.1%.
2) The body has biological repair mechanisms that come into play when small-to-moderate levels of harmful DNA damage are produced. Therefore, the LNT risk model overestimates the fatal cancer probability at these low dose levels by quite a large amount, and this becomes larger and larger as the radiation dose level decreases and the repair mechanisms can do a more effective job. In fact, there is experimental evidence that at the radiation dose levels we are discussing here (i.e., 0 - 10 rem, or 0 – 0.1 Sievert), radiation and other carcinogens can sometimes cause a slight decrease in the fatal cancer incidence. This mechanism, which is not yet fully understood, is called “radiation hormesis”. Radiation hormesis is still in the closet among many practitioners of radiobiology, but more and more evidence to support radiation hormesis is gradually emerging, to the extent that sometime in the near future I believe the LNT model will be abandoned for the typical diagnostic radiology dose range of 0 - 10 rem (0 - 0.1 Sievert) in favor of the “hormetic dual-probability model”, which predicts a negative fatal cancer risk within this dose range. It has been a standard medical practice in Europe for centuries to expose patients to high concentrations of radon gas present in certain geological regions with the intent of strengthening their immune systems and hence making them better able to combat various diseases they may have.
3) The use of the LNT risk model has further limitations. A straight line through a very scattered array of data points often belies large statistical uncertainties in the slope of that line. Such uncertainties may hide fine structure of the radiation dose–fatal cancer incidence relationship that are not clearly evident. One such proven departure from the LNT relationship is the existence of a threshold for fatal cancer risk. In this modified model, a certain amount of radiation exposure can be tolerated without any detectable rise in the fatal cancer rate. Only after the radiation dose level exceeds a so-called dose threshold, does the fatal cancer rate start to climb in a more conventional straight-line fashion. This phenomenon of dose/effect threshold has been studied extensively, and despite the scattered nature of the experimental data some cancers have been found to conform better to the threshold version of the LNT model than to the non-threshold linear LNT model.
It is sometimes instructive to compare the lifetime risk of fatal cancer resulting from radiation exposure with the risks of death due to non-radiation-related factors. The table below shows such a comparison.
Approximate Lifetime Risk of Death Due to Receiving 1 Typical CT Scan vs. Various Non-Radiation Risk Factors.
1) The majority of radiation dose received by the general population is due to Diagnostic x-ray procedures and naturally occurring background radiation.
2) At the upper end of the dose range of single diagnostic radiology procedures (for example, a single abdominal or pelvic CT scan), the probability of induction of a fatal cancer is very roughly 0.05%.
3) With exposure only to natural background radiation, a 50 year-old individual (having accumulated approximately 300 orem/year of effective dose during each year of life) would have very roughly a 0.7% probability of contracting a fatal radiation-induced cancer.
4) The baseline lifetime fatal cancer risk in the population due to all causes is approximately 20%.
5) Our simple calculations are based on the LNT cancer risk model. Although this model pretty much utilizes the best radiation effects data we have at the present time, it has very large associated statistical uncertainties—often as large as +100%. Additionally, our calculations ignore the possible presence of a radiation effect threshold or of the protective effects of biological DNA repair mechanisms. The collective impact of these omissions in the basic LNT radiation risk model probably results in a large overestimate of the fatal cancer risk, especially at low diagnostic x-ray dose levels.
6) There is significant experimental evidence that at the radiation dose levels we are discussing here (i.e., below 10 rem, or 100 mSv), radiation can in fact cause a slight decrease in the cancer rate. This is termed radiation hormesis, and the associated dual-probability hormetic model that uses this concept will most likely soon replace the “conservative” LNT model on which all cancer inducing radiation effects in humans are traditionally based.
About the Author