In his job as a radiation oncologist, Jim Welsh has treated hundreds of people with cancer by exposing them to just-toxic-enough doses of radiation. But in the last five years, his patients in remission have started asking some disconcerting questions. They would come back every once in a while, every few months or years, to get a new scan, to make sure that their cancer hadn’t returned. But they were hesitant. They wanted to know if the radiation used to treat them and take those follow-up images could actually cause new tumors.
In the past decade, patients and doctors have become more careful about how they use medical radiation to diagnose and treat disease. “Fifteen years ago, nobody was asking this question,” Welsh recalls. “But when people realized just how much of an increase there was in the use of medical radiation diagnostically, they started wondering if we’ve also caused a lot of cancer along the way.” That realization sparked nationwide and local initiatives to cut back on medical radiation, especially for procedures like CT scans, which require high doses to generate those lovely, accurate pictures of bones and innards. But Welsh and a minority of radiation experts think that the medical community might be overcorrecting—that concerns like Welsh kept hearing from his cancer survivors are keeping them from getting crucial tests to keep them healthy.
That contradiction stems from one big problem: It’s nearly impossible to know whether low doses of radiation cause cancer. High doses are a problem, obviously; the best data for that comes from long-term studies of the survivors of nuclear disasters like the atomic bombs dropped on Hiroshima and Nagasaki 70 years ago, and the Chernobyl nuclear power plant disaster in 1986. In those accidents, people were exposed to 200 milliSieverts of radiation, on average (Cancer therapies often use doses in that range, too). A group called the Radiation Effects Research Foundation still follows the medical outcomes of Hiroshima and Nagasaki survivors.
For outcomes from low doses, though, researchers extrapolate backwards from those data points. That’s how medical researchers have modeled the effects of medical radiation: the fraction of a mSv in a typical X-ray, or the 5 to 10 mSv in a standard CT scan. They use what’s called a linear no-threshold model, which draws a straight line (thus the linear) from the high-dose cancer risk data points down until zero exposure equals zero risk. “The idea of the LNT is to understand the harmful effects in the very low-dose range, around 1 mSv, where we don’t have direct epidemiological evidence,” says Rebecca Smith-Bindman, a radiologist and epidemiologist at UC San Francisco. In other words, the model says there’s no safe dose of radiation.
But some researchers say that’s bunk. “At the doses used in medical imaging, there is no evidence that radiation causes an increase in long-term risk,” says Cynthia McCollough, director of the CT Clinical Innovation Center at the Mayo Clinic. “There are models and hypotheses that it might, as well as models and hypotheses that low-dose radiation is actually beneficial.”
McCollough may be right on that second point—models like the LNT can show whatever theorists want them to show. Naturally, epidemiologists would like to supplant the models with real data. And researchers have tried: Plenty of studies have shown a small association between radiation and cancer risk. But plenty have failed to show an association.
On top of that, whenever researchers study medical radiation, they run into confounding variables. A person who wanders into an ER, gets a CT scan, and 10 years later gets cancer is a data point—a correlation between radiological imaging and cancer incidence. But “any observed association may be due to the medical symptoms that led to the imaging exam, and not the exam itself,” McCollough says.
That means data from Hiroshima and Nagasaki survivors—including those who were exposed to the bombs’ radiation while still in the womb—is often the best researchers can get. As those survivors age, they’ll continue to be monitored for cancer developments, and any correlations between their exposure levels and cancer incidence will become more robust. Smith-Bindman says that the most recent follow-ups of survivors who were young or in utero at the time of the detonations actually indicate a higher cancer risk than earlier versions of the LNT model suggest.
But a better long-term strategy will be to compare apples to apples: accumulating data about cancer in enough patients exposed to medical radiation that epidemiologists can discount the influence of background cancer rates. So far, the most notable study to do that, published in the British medical journal the Lancet in 2012, looked at the risk of leukemia and brain tumors in more than 175,000 children exposed to relatively high levels of radiation from CT scans. It projected that every 10,000 head CT scans would result in about one extra case of each cancer.
That trial was retrospective—looking backwards into medical records so the researchers would have enough cancer incidences to pull out data. The next step, being taken by researchers like Smith-Bindman, is to study the effects in a population going forward, a prospective study. One $10 million study will track 7 million children over 18 years to assess their radiation exposure and cancer incidence.
Until then? “It’s a very legitimate question, and it’s our job to inform our patients that nothing in medicine is a free ride,” Smith-Bindman says. “You don’t want to have an unreasonable fear of radiation; the fear is that it’ll cause irrational concern and patients will decline what we think is beneficial treatment.” Every medical procedure—whether it involves radiation or not—comes with a certain risk of harm. The trick is figuring out whether it’s worth it