Strange Glow

by Timothy J Jorgensen

  Despite the good fortune of having multiple cohort studies to work with—an epidemiologist’s equivalent of striking gold—the analytical efforts of the expert panels were hampered. This is because most of the miners were also smokers, and smoking produces far more lung cancer than radon. Teasing apart the relative contributions of smoking versus radon to the lung cancer incidence in miners has been the greatest challenge to radon epidemiologists. This is not only because smoking was so prevalent among miners and nonminers alike, but also because smoking makes the lung cancer risk from radon worse—much worse. Smokers, therefore, represent a subset of the population that is particularly sensitive to radon.

  Identifying sensitive subpopulations who may require special protection is an important task of any risk assessment process. Standards that protect the average person are inadequate for highly sensitive individuals. For example, pollen counts that would not bother most people might trigger anaphylactic shock in someone highly allergic to pollen. Sensitive subpopulations have a major influence when it comes to setting exposure limits for the toxins and carcinogens in our environment, because the limits chosen by regulators must protect everyone, even the most sensitive. In practical terms, this means that exposure limits set to provide sufficient protection for sensitive subpopulations amount to overprotection for people with normal sensitivity. When setting limits for radon, the subpopulation needing the most protection is the smokers, and it’s the nonsmokers who enjoy overprotection.

  It’s not clear why smokers are hypersensitive to radon. It could be because radon and the chemicals in cigarette smoke damage DNA in somewhat different ways. For example, the ionizing radiation from radon tends to break DNA, while the highly reactive chemicals in cigarette smoke (e.g., benzo[a]pyrene) tend to attach themselves to the DNA and form what chemists call bulky adducts (because they represent large chemical additions to the DNA structure). These different forms of DNA damage may impede each other’s repair, such that DNA breaks block the bulky adduct repair processes, and bulky adducts block the repair processes of the breaks. Thus, having both types of DNA lesions at the same time might be much worse that having one type alone.

  Alternatively, smokers may be hypersensitive to radon because they have damaged their lungs to the point that their bronchi—the treelike tubes of air passageways—are no longer able to efficiently remove air particles that are breathed in. Radon often arrives in the lung attached to dust and other types of particles in the air. A nonsmoker’s lungs are highly efficient at expelling particles up from the lungs and into the throat, where they subsequently get swallowed along with saliva and exit the body in feces. Smokers, however, slowly lose the cells lining the bronchi that perform this function, so particles tend to stay trapped in the lungs. In the case of radon, it could mean that particles laden with radon may stay in the smoker’s lungs longer, resulting in a higher radiation dose to a smoker’s lungs than a nonsmoker’s lungs, even when they’re breathing the same amount of radon.

  It’s not clear which of these two possible mechanisms is correct. It might even be both, or perhaps neither. There is no strong evidence that can explain exactly why smokers are hypersensitive to radon. Nonetheless, they definitely are at higher risk of lung cancer caused by radon—six to eight times higher.

  HABER’S RULE

  In order to set limits to radiation dose, you need to be able to measure it. For most types of radiation exposure, measuring dose is relatively straightforward. Since radon is a gas, however, measuring radiation dose to the lung is not that simple. It can be done, but relies on the use of mathematical models along with some assumptions about lung physiology. This is analogous to the bone physiology assumptions used by Robley Evans to calculate the incorporation of radium into the bones of the unfortunate radium girls. Also, lung physiology is very complicated and many of the parameters required to model dose cannot be directly measured, only estimated. The combination of multiple physiological assumptions and multiple estimated parameters increases the uncertainties about modeled radon lung doses to the point that confidence in their accuracy is compromised. Without reliable dose estimates, risk assessment falls apart. Fortunately, another strategy is available.

  The alternative approach is to make use of a simple mathematical relationship known as Haber’s rule for gaseous hazards:

  D = C × E

  where D is dose, C is concentration, and E is exposure time. Haber’s rule tells us that dose is approximately proportional to both the concentration of the noxious agent in the air and the duration of exposure to that agent. More scholarly dosimetric models may give more precise dose estimates, but Haber’s rule seems to be remarkably accurate in predicting health risks from gases, without any of the complicated statistical hoopla that surrounds more erudite dose modeling.8

  Haber’s rule tells us something else. As we’ve learned, dose drives biological damage, so dose is what we want to limit. How can we do this? Well, in situations where we have limited control over radon concentration (i.e., in mines), we can measure the concentration in the mine air and then limit how long the miners are allowed to work in that particular concentration. In situations where we have no control over exposure time (e.g., the duration of a family’s occupancy of a radon-contaminated home), we can achieve the same end by just assuming maximal occupancy time in the house as a worst-case scenario, and then limit the radon concentrations in the house to ensure that the dose limit will not be exceeded. Thus, controlling the concentration of radon in the air of homes is a useful means to limit he radiation dose to the inhabitants. This is the approach the EPA has taken to protect the public from the residential radon threat.

  For all of the above reasons, radon is one of the few radiation hazards not typically measured in millisieverts (mSv). Rather, Haber’s rule has been invoked to provide a radiation dose unit unique to radon. Since most of the data comes from mine workers, the experts have chosen to calculate lung cancer risk from radon using a unit called the working level month (WLM). Although it has a technical and precise definition, for our purposes it can just be defined as the radon dose that a typical miner might receive from working in a typical radon-contaminated mine for one month.9 Simply stated, it is just the radon concentration of the mine, expressed in working levels (WL), multiplied by the exposure time measured in months (M), à la Haber’s rule. To be perfectly accurate, it’s a measure of exposure, not dose. (As we know, ionizing radiation dose is measured in mSv.) But WLMs should be proportional to the true dose, so it can be used as a surrogate for the actual dose.

  Expressing radon dose in WLMs also has another practical advantage. It allows us to depict other radon exposures, such as living in a radon-contaminated home, in terms of the equivalent mining work it would represent.

  GETTING THE FACTS STRAIGHT: LNT

  Because “the dose makes the poison” (remember Paracelsus), we need dose-response data to identify the dose where radon levels become a health concern. Since lung cancer is the health effect of interest, “doses” (i.e., WLMs) are typically plotted against lung cancer incidence rates, and fitted to a straight line. The fitted line must be drawn straight, rather than curved, because radon epidemiologists use the same assumption about the cancer risk from internal radon exposure as the atomic bomb epidemiologists use for estimating cancer risk from external exposure: Cancer risk is directly proportional to radiation dose at all dose levels, no matter how low those doses are. Epidemiologists call this the linear no threshold (LNT) model of cancer risk assessment. Adopting this linear, dose-response assumption for radon provides consistency across both internal and external radiation risk assessments. And consistency is always good policy for risk assessment.

  The LNT model also has a mechanistic justification. Unlike the radiation sicknesses that we learned about earlier, where a certain threshold number of cells must be killed before health effects can be seen, scientists think that cancer induction behaves differently. Most scientists believe that there is a finite probability of con
tracting cancer from any exposure to a carcinogen because there is no dose so low that DNA damage does not occur, and DNA damage is the precursor to cancer. Thus, conceptually at least, even one interaction of radiation with a single cell’s DNA could cause that cell to become a cancer cell. The probability of that happening is admittedly extremely low, but it is not nonexistent. Consequently, epidemiologists reject the concept of a threshold when dealing with carcinogens, and just assume risk is proportional to dose in a straight-line (i.e., linear) relationship, regardless of how low the dose is (i.e., they assume there is no threshold).

  As we mentioned earlier when we discussed the cancer risk calculations generated from atomic bomb victim data, not all scientists favor linear models. Some strongly believe that the LNT approach to radiation protection has exaggerated the risk of cancer from low radiation doses, and their arguments have some validity.10 In contrast, no credible scientists believe that the LNT risk assessment approach underestimates low-dose risk. Thus, the LNT model is generally considered the most conservative and, therefore, the most responsible way to determine cancer risk, precisely because it assumes the highest theoretical risks at the lower doses. Since such low-dose risk estimates are at the high end of reasonable predictions, dose limits based on LNT models afford the greatest degree of protection to the public. For this reason, the dogma among public health professionals is to use LNT as the default model for assessing cancer risk for radiation, and to resist any divergence from LNT modeling unless there is overwhelming evidence that it is inapplicable—a situation that seldom exists.11

  CUTTING TO THE CHASE

  We’ve taken the long route to get to the final radon risk estimates, but our incursion into the underlying risk assessment methods was important. Understanding the methods and approaches used in risk assessment does three things. It allows us to appreciate the strengths of the risk estimates, to recognize their weaknesses, and to gauge our level of confidence in their accuracy. Now that we know how radon risk estimates are generated, let’s take a look at the final numbers.

  The scientific experts’ best estimate of the lung cancer risk per WLM is approximately 0.097% for smokers, and 0.017% for nonsmokers.12 Stated simply, working for one month in a typical uranium mine increases the odds of getting lung cancer by about 1 in 1,000 for a smoking miner, and 1 in 6,000 for a nonsmoking miner.

  Another way of expressing the risk of one month of work is as follows: For every 10,000 smoking miners that worked in a mine for one month, we would expect 10 of them to eventually contract lung cancer from their radon exposure. Alternatively, if 10,000 nonsmoking miners worked for one month, we would expect fewer than 2 of them (i.e., 1.7 miners) to get lung cancer. As you can see, the smokers have a much higher cancer risk from radon. Many people find this way of depicting risk to be intuitively comprehensible because it allows them to immediately see how frequently the bad outcome happens to a group of exposed versus unexposed people.13

  MI CASA ES SU CASA

  All right you say, so much for the mines. What about my home? Here’s the EPA’s logic on home radon risks. The EPA has decided that a home resident’s radon dose should be no more than 2% of a typical mineworker’s. That means the concentration of radon in home air should be at or below 0.02 WL (4 pCi/L; 150 Bq/m3). To go much lower than 0.02 WL would be technically difficult (i.e., extremely expensive) and practically unwarranted, given that even outside air has an average radon concentration of 0.0025 WL.

  What is the risk level to residents living their entire life in a home with radon concentrations at 0.02 WL? According to the EPA’s calculations, if a person lived in that home every day of his life, occupied the house for up to 17 hours per day (70% of the time), and lived to be 75 years old, then that person’s lifetime risk of lung cancer from the radon in his house would be 6.2% if he were a smoker, but only 0.73%, if he were a nonsmoker.14 But exactly how bad are these levels of risk? Let’s explore this question.

  NUMBER NEEDED TO HARM

  Instead of starting with 10,000 people and tracking their health outcomes as we did above for the miners, another way to portray risk is simply to ask the following question: “How many people would need to receive this particular dose level before one of them would be expected to have a harmful outcome (e.g., lung cancer)?” In other words, what number of people need to be exposed in order to harm just one? When risk estimates are framed around this question, the risk metric is called the number needed to harm (NNH).15 It is an underutilized metric for risk characterization, but can be highly valuable because it gives a vivid sense of the magnitude of the risk and allows easy side-by-side comparisons of risk levels for various exposures. As an example of how NNH values work, let’s use the approach to characterize the lung cancer risk in a home with radon levels at the EPA’s exposure limit.

  Restating the lung cancer risks from lifetime residence in a home at the EPA radon limit in terms of NNH units yields the values of 16 and 137 for smokers and nonsmokers, respectively. This means that out of 16 smoking lifelong residents of such radon-containing houses, one would be expected to develop lung cancer from the exposure, while it would take as many as 137 nonsmoking lifelong residents for one to become afflicted with lung cancer. As you can see, the lower the NNH, the greater the risk; so it’s much more dangerous for a smoker to live in the radon-contaminated house than the nonsmoker.

  It’s important to remember that there’s nothing magical about the NNH. It’s just one of many ways to statistically depict risk. But studies have shown that characterization of risks as an NNH makes complicated risk scenarios comprehensible to people from a variety of backgrounds, and helps them develop a more accurate and realistic understanding of their personal level of risk.16

  We’ll see more uses for the NNH and its mirror-image twin, the number needed to treat (NNT), which is a metric of benefit that is sometimes used to assess the effectiveness of various medical treatments. We’ll also learn how the NNH and the NNT can be considered together to weigh the risks and benefits of a particular type of radiation diagnostic procedure. We’ll even learn how to calculate the NNH and NNT on our own when the risk assessors fail to provide them to us. But for now, let’s just start thinking in terms of NNHs and NNTs as a way to clear our minds of the confusion that other risk metrics, such as odds ratios and relative risks, can often cause.17 Confusion is not a good starting point for making valid risk decisions.

  ALL TOO COMMON

  If these radon risk levels still seem high to you, consider this. The chances of a nonsmoker contracting a fatal cancer of any type during her lifetime is approximately 25%, a number that is unfortunately high. The additional risk from living in the house with radon at the residential limit, according to the numbers quoted above, would increase that risk from 25% to 25.73%. But smokers, in contrast, already have an elevated baseline risk of developing a fatal cancer approaching 50% (depending upon how much they smoke), apart from any radon exposure they might have. For a smoker, the risk of developing a fatal cancer during a lifetime thus moves from about 50% to 56.2%.

  Also, consider the fact that virtually no one matches the EPA’s extreme exposure assumptions. Rather than living in a single home for an entire lifetime, typical Americans have more than 10 different residences during their lifetime. Since radon contamination of homes is rare, the chances that more than one of these residences would have high radon levels are fairly remote. Also, the prospect that a person would spend as much as 70% of his 75-year lifetime inside any single house is also remote. Thus, in all likelihood, no one in the United States living in a radon-contaminated house fits the EPA’s worst-case description for residential radon dose (i.e., 70% of a 75-year lifetime occupying one house). Thus, the actual doses to residents living in houses at the EPA exposure limit of 0.02 WL are probably less than a tenth of the theoretical worst-case doses and, likewise, their lung cancer risk would also be less than a tenth of the risk levels calculated above.

  WHO’S DOING THE DYING?

 
Based on the EPA’s risk assessment numbers and their underlying assumptions, the agency has claimed that high levels of radon in homes theoretically kill as many as 21,000 Americans every year.18 This amounts to 13% of the 160,000 annual lung cancer deaths in the United States, and 3.5% of the 585,000 total annual cancer deaths of all types.19 But who are these “theoretically” dead people anyway?

  Remember that smokers are six to eight times as likely to get radon-induced lung cancer, and consider that the prevalence of smokers in the US population is about 18%.20 This large difference in sensitivity between smokers and nonsmokers, coupled with the fact that smokers are fairly common within the American population, allows us to make yet another prediction: Most victims of radon-induced lung cancers will be current smokers or former smokers. This fact, which expert panels have stressed time and again, greatly influences an individual’s personal risk of getting lung cancer from radon.

  Death from radon-induced lung cancer is actually quite rare among nonsmokers. Even if you accept the EPA’s estimate that 21,000 annual lung cancer deaths in the United States are due to radon exposure in homes, about 19,000 of those dead would be smokers. Only 2,000 deaths would occur among nonsmokers, just 0.3% of the total annual cancer deaths in the United States. The 2,000 deaths per annum would put the rate of radon-induced lung cancer among nonsmokers at about one tenth of the annual deaths from influenza.21 In short, you must be very unlucky indeed to die from radon-induced lung cancer if you’re not a smoker.

  THE EPA CONTROVERSY

  When the EPA first issued its radon guidelines in the 1980s, the public was slow to comply. The same public that was in a panic over asbestos and formaldehyde contamination of their homes seemed unperturbed about radon. Social psychologists say that is likely because radon is a natural hazard, as opposed to a manmade hazard.22 For whatever reason, people seem to be less afraid of natural hazards than manmade ones. Regardless of the explanation for the blasé attitude, the EPA warnings were not being heeded. So the EPA increased its hype with an advertising campaign that focused on the extreme situations and downplayed, and even concealed, that the lung cancer risk of radon was largely restricted to smokers.23