About 40% of cancers among Americans can be attributed to potentially modifiable factors such as smoking, drinking, obesity, and physical inactivity. If a widely reported study from earlier this year is right, routine CT scans belong on that list, too — near the top, in fact.

The study, published in April in JAMA Internal Medicine, estimated that CT scans performed in the US in 2023 alone — 93 million among 62 million patients — will result in over 100,000 lifetime cancers. Framed differently, the authors concluded, CT scans could eventually account for roughly 5% of all cancers diagnosed among Americans each year. That would put the exams on par with obesity or overweight (7.6% of all cancers) and alcohol consumption (5.4%), which along with cigarette smoking (19.3%) constitute the three leading preventable causes of cancer.

“It’s a big number,” lead author Rebecca Smith-Bindman, MD, professor of epidemiology and biostatistics, University of California San Francisco, told Medscape Medical News.

The study is one of the several published in the past couple years raising concerns about the cancer risks associated with CT scans, mainly in children and teenagers. The most recent, which Smith-Bindman’s team published earlier this month in The New England Journal of Medicine (NEJM), estimated that 10% of pediatric hematologic cancers may be attributable to lower-dose radiation from medical imaging, particularly CT scans.

To Smith-Bindman, the evidence underscores the long-recognized need to limit low-value CT scans and supports efforts to reduce patients’ radiation exposure during necessary ones.

However, experts have urged caution in interpreting these findings.

The JAMA study’s big number, for instance, is based on a modeling exercise, and not everyone agrees that it necessarily reflects reality.

The takeaway from the analysis is that the 100,000+ estimate is almost certainly too high, according to Gideon Meyerowitz-Katz, MPH, PhD, an epidemiologist and senior research fellow at the University of Wollongong in Wollongong, Australia.

As for the 10% figure from the NEJM study, Lindsay M. Morton, PhD, characterized the risk for pediatric hematologic cancers to be “small.” Among children exposed to cumulative radiation doses of 30 mGy or more, the excess incidence of those cancers was 25.6 per 10,000 — but less than 1% of the study population was exposed to doses that high.

Ultimately, it’s key for all clinicians involved in medical imaging to “wisely interpret” the results of these studies “to understand the balance of the very small risks and the notable benefits of necessary imaging examinations,” Morton, of the National Cancer Institute (NCI) in Bethesda, Maryland, wrote in an editorial accompanying the NEJM study.

What Is the Actual Risk?

CT scans use ionizing radiation, a known carcinogen.

“So the theoretical risk [to patients] is something we’ve grappled with since the advent of CT,” said David B. Larson, MD, chair of the American College of Radiology’s Commission on Quality and Safety and professor of radiology at Stanford University, Stanford, California.

And Smith-Bindman’s studies are not the first to try to put hard numbers on that theoretical risk. In 2009, an NCI study estimated that CT scans could be responsible for 29,000 cancers annually or 2% of all cancers among Americans.

The more recent projection in JAMA is substantially higher for a few reasons, Smith-Bindman said. One is that CT use, including low-value testing, is up 30% since the earlier study.

More importantly, the new estimate also assumes greater radiation exposure per exam: Smith-Bindman’s team calculated organ radiation doses based on data from over 120,000 CT scans, whereas the NCI study relied on imaging protocols and national survey data.

The latest study also accounted for the common practice of multiphase scanning, which involves taking images at multiple timepoints as the contrast dye moves through patients’ veins and arteries. A 2023 study from Smith-Bindman found that multiphase scanning was the major driver of radiation dose variation during routine abdominal CT scans.

For the JAMA study, “we had more information about what the doses really are in practice,” Smith-Bindman said.

However, critics say that this additional information does not necessarily paint a more accurate picture of the cancer risks to patients.

“I think the key challenge with this study is that there are so many assumptions in the model,” Meyerowitz-Katz told Medscape Medical News.

“The main reason the numbers are bigger in this study,” he said, “is that the authors have assumed that radiation doses are much higher [now]. And while there is some evidence of this, there are still places where they’ve extrapolated without a huge quantity of data.”

Critiques from two letters, published earlier this month in JAMA Internal Medicine, urged caution in interpreting the study’s estimate, also citing overestimates of radiation doses. One of the letters points out that the projections do not account for recent CT technology advances that may reduce patients’ radiation exposure now and in the future — including artificial intelligence (AI)-based imaging software and photon-counting CT scans.

“This leads to an overestimation of the projected risk of cancer alongside the upward trend of CT use,” wrote Chris Hsu and colleagues at University of Texas Medical Branch, Galveston, Texas.

Another issue, Meyerowitz-Katz said, is the tool that Smith-Bindman and her team used to make its calculations — an NCI calculator called RadRAT — is largely based on cancer cases among Japanese atomic bomb survivors. There are questions about how well it can be applied to estimating the risks of exposure to low-dose ionizing radiation from medical imaging.

The tool assumes that people exposed to radiation have an average life expectancy for their age and sex, but that’s likely untrue for many people undergoing CT scans whose cancer risk could be extremely variable, Meyerowitz-Katz explained.

The study authors also pointed out this limitation, noting that “the degree to which patients who undergo CT have shorter life expectancy due to underlying illness may overestimate future cancer risk.”

Richard Wakeford, PhD, honorary professor of epidemiology, The University of Manchester, Manchester, England, agreed that “considerable caution” is needed in using RadRAT to predict excess cancers from CT scans.

“This is largely because of the substantial assumptions that must be made in applying risk models derived from epidemiological studies of populations briefly exposed to moderate and high doses — primarily the Japanese survivors of the atomic bombings of Hiroshima and Nagasaki — to low-level exposure circumstances,” Wakeford said.

A Better Estimate?

Smith-Bindman pointed to yet another criticism she’s heard since the study’s publication: the lack of “direct evidence” from these findings, or otherwise, that CT scans are harmful.

“That’s like saying there’s no direct evidence that smoking causes lung cancer,” she said. “CT is associated with cancer risk. Just like everything else in healthcare, it has potential benefits and potential harms.”

Meyerowitz-Katz agreed that CT scans may carry some cancer risk for patients but said a modeling study is not the best way to define it. Better evidence would come from cohort studies looking at long-term outcomes of adult patients with the same medical condition who either underwent CT scans or didn’t.

Much of the existing evidence on CT exposure and cancer risk comes from studies evaluating pediatric populations.

Smith-Bindman’s recent NEJM study, for instance, was a retrospective cohort study, which used medical imaging history data of 3.7 million North American children born between 1996 and 2016 to estimate the radiation doses to active bone marrow. The researchers concluded that among children exposed to doses of at least 30 mGy, the excess cumulative incidence of hematologic cancers by age 21 was 25.6 per 10,000.

A previous study, conducted in nine European countries using similar methods, reached similar conclusions. A case-control study from Finland published earlier this year found a heightened risk for brain tumors among children and adolescents who underwent one or more head or neck CT scans.

However, studies of adults, looking at specific cancers such as breast cancer, meningioma, and thyroid and hematologic cancers, have not necessarily found a significant link between CT scans and cancer risk. The breast cancer study, for instance, found that women exposed to thoracic CT scans during or after pregnancy did not appear to have an increased risk for maternal breast cancer in the short term.

Although cohort studies are “prey to all sorts of biases,” Meyerowitz-Katz acknowledged, these studies “would, I think, give us a better idea than extrapolating from nuclear bomb survivors to the entire population of the US.”

For now, Wakeford told Medscape Medical News, “it should be assumed by physicians that some small risk of cancer will arise from radiation exposure during a CT scan.”

While that should be a consideration when ordering exams, Wakeford said, “I would hope that it would form (only) a small part of that clinical judgement.”

Implications for Practice

Regardless of the accuracy of the estimates, Larson said the findings are a reminder to physicians to use CT scans judiciously. In practice, he told Medscape Medical News,it’s impossible to calculate any individual patient’s cancer risk from CT scans. But if that exam is appropriate, Larson said, patients can be assured that “the benefit of the information you’re getting is much greater than the theoretical risk.”

The problem, Smith-Bindman said, is that CT scans have long been overused to detect suspected conditions ranging from kidney stones to pulmonary embolism, and efforts to curtail overuse, like the Image Wisely campaign, have not moved the needle much.

She sees a greater opportunity for reducing patients’ unnecessary radiation exposure by working on dose optimization. In a 2019 study of 151 institutions across seven countries, Smith-Bindman and her colleagues found that mean effective radiation doses and the proportion of high-dose CT scans varied hugely among countries. For example, the proportion of high-dose abdominal CT scans (radiation doses above the 75th percentile) ranged from 4% in Netherlands to 69% in Japan, with the US sitting toward the middle at 22%. Another 2019 study by her team found that across US hospitals, radiation doses used for lung cancer screening varied, often surpassing American College of Radiology guidelines.

“I think we need standards across our field,” Smith-Bindman said. “If I’m a patient, I should be assured that the radiation dose for a given CT exam is going to be the same no matter which hospital I go to. A radiologist’s preferences for image quality shouldn’t be the driver.”

Larson agreed that “there’s definitely room for improvement in dose optimization and process controls at almost every institution.”

Exactly how to achieve that is another matter. Having the latest or at least more modern CT equipment helps, Larson said, because this newer tech is more dose efficient. Software solutions, using AI or iterative reconstruction, may also reduce radiation doses, he added.

One national effort to cap CT radiation doses took effect this year: a new quality measure adopted by the Centers for Medicare & Medicaid Services (CMS) into pay-for-performance programs. Smith-Bindman, who led the development of the measure, said it holds hospitals and doctors accountable to using evidence-based radiation doses during CT scans.

However, the measure faced pushback from the American Association of Physicists in Medicine, and CMS recently proposed a major modification: Maintain indefinite voluntary reporting on the measure rather than making it mandatory beginning in 2027.

According to Smith-Bindman, that would “gut the only guardrails” that currently exist for standardizing CT radiation doses.

Regardless of what happens with the CMS proposal, Smith-Bindman said there are also steps that radiologists “can take immediately” to limit patients’ radiation exposure without compromising image quality — including questioning the use of multiphase CT scans.

While multiphase techniques produce “phenomenal” images, they do not necessarily aid in diagnosis, Smith-Bindman said. “Our focus has not been, how can we get diagnostic information with the lowest radiation dose possible?” she said. “I think that’s what we need to do. As a field we need to say this is a priority.”

Smith-Bindman reported being a co-founder of Alara Imaging, a company focused on improving the clinical and operational aspects of health systems. The CMS quality measure uses intellectual property she developed. Wakeford reported having been a member of expert committees addressing radiation risks, such as the International Commission on Radiological Protection and the United Nations Scientific Committee on the Effects of Atomic Radiation. Larson reported receiving research support from Siemens Healthineers.