CT scans and cancer risk

“CTs could raise the risk of cancer,” The Independent reported. It said that as many as one in 80 people could be at risk of developing cancer as a result of having a computed tomography (CT) scan.

The report is based on two studies that estimated the future risk of cancer from CT scans for people in the US. The figures are estimates only and are based on data from various sources, which may result in some inaccuracy. Also, the results cannot be generalised outside the US. This includes the UK, where CT scans may not be used as frequently.

It should be emphasised that people who have CT scans are likely to be exposed to a very small individual risk. These studies are calling attention to the issue that when more people are exposed to radiation from CT scans, the collective risk increases, and more cancer cases can be expected. These findings highlight the need for clinicians to weigh the risk of radiation exposure from a scan against its benefits.

Where did the story come from?

The Archives of Internal Medicine has published a series of articles on the risks from radiation exposure from CT scanners, including a modelling study, a cross-sectional study and an editorial discussing the issue.

The research for the modelling study was carried out by Dr Amy Berrington de Gonzalez of the National Cancer Institute, Bethesda, Maryland, and colleagues from other institutions in the US and Korea. The cross-sectional study was performed by Dr Rebecca Smith-Bindman from the University of California and other institutions in the US. The editorial was written by Dr Rita F Redberg. This is a top level appraisal of the research published in the two scientific articles.

The modelling study received an author grant from Siemens Medical Systems. The cross-sectional study was funded by the National Institutes of Health (NIH), National Institute of Biomedical Imaging and BioEngineering, National Cancer Institute and UCSF School of Medicine Bridge Funding Program.

What kind of research was this?

The use of CT scans in the US has reportedly tripled since 1993 to its present level of about 70 million scans per year. While these tests have proved to be of great value in the diagnosis of disease, the possible risks from exposure to radiation have caused some concern. The two studies reported here investigated this issue.

The first was a modelling study designed to estimate future cancer risks from CT scan use in the US, with risks assessed separately for different ages, sexes and scan types. This study used various data sources to calculate risk estimates and predict the number of cancers expected due to radiation.

The second study was a cross-sectional study investigating the radiation doses typically received from CT scans. Although CT scanning involves higher doses than conventional X-rays, typical doses are not known. These researchers aimed to estimate the radiation exposure from CT scanning, and to quantify the potential associated cancer risk.

Both of these studies involved predictions and estimates of numbers of cancers associated with CT. Although both studies used the best resources available to them, there may have been some unavoidable inaccuracy or imprecision in the estimates.

What did the research involve?

Modelling study
The modelling study used data from previous research to estimate the cancer risk of each scan type to specific groups, and the average number of radiation-related cancers that would develop. A cancer-risk projection model was then calculated for the US population using the following resources.

• The estimate of the frequency and type of scan performed in 2007 was calculated from Medicare claims and the IMV Medical Information Division survey of CT scan use.
• Organ-specific radiation received by age and sex was gathered from national surveys.
• The researchers also used The National Research Council’s Biological Effects of Ionizing Radiation (BEIR) report in their calculations, which is a comprehensive review of the health risks from low-level radiation. The researchers made minor modifications to the risk models in this report and developed additional models for areas that were not covered.

Cross-sectional study

The cross-sectional study looked into the radiation doses associated with the 11 most common types of CT scan. To find the 11 most common scans, the researchers used data from one month (March 2008) from the UCSF Radiology Information System, which contains information on all CT scans performed in the US.

The researchers then looked specifically at CT scans of 1,119 consecutive adult patients at four hospitals in California between January and May 2008. Scans performed for treatment purposes (for example CT-guided abscess drainage) were excluded.

They compared radiation doses for these CT procedures with those of other investigations such as X-ray and mammography. To estimate the cancer risk from CT scans at different doses, they used the methods given in the BEIR report to estimate the lifetime attributable risk (LAR) of cancer. This is defined as additional cancer risk above and beyond that which any person would normally have, and is a measure of how many additional years of life might be gained by removing the radiation.

Both studies used complex risk models and data from reliable sources to calculate cancer risk and the average level of radiation exposure by age and sex. Although the researchers used the best data available to them, these are only detailed estimates and cannot be considered definite risk figures. There is likely to be some inaccuracy from the different data sources used and because different types of radiation exposure are included.

What were the basic results?

The modelling study estimated that, on average, 29,000 future cancers in the US could be related to CT scans performed in 2007. The largest contributors were calculated to be scans of the abdomen and pelvis (14,000 cancers), chest (4,100) and head (4,000), as well as scans where high-dose radiation was used. A third of the expected cancers were attributed to scans performed between the ages of 35 and 54 years, while 15% were attributed to scans in those under 18 years. Two-thirds of the CT-related cancers were expected to be in women, due to the larger number of CT scans carried out in women.

In the cross-sectional study, the average age of patients when scanned by CT was 59 years, and 48% of the patients were women. The 11 most common types of CT scan made up about 80% of all CTs performed. Radiation doses varied significantly between the different types of CT scan, with average doses ranging from 2 millisieverts (mSv) for a routine head CT to 31mSv for a multiphase abdomen and pelvis CT scan. Doses also varied within and between hospitals, with an average 13-fold variation between the highest and lowest dose for each scan type. The estimated number of CT scans that would lead to the development of a cancer varied depending on the type of CT and the patient’s age and sex.

It was estimated that one in 270 women who had CT coronary angiography (a fairly high-radiation-dose scan of the heart blood vessels) at the age of 40 would develop additional cancers from that CT scan (one in 600 men) compared with an estimated one in 8,100 extra women who had a routine CT scan of the head (one in 11,080 men). Risk for developing cancer in later life was higher for a person scanned at a young age, and lower for a person scanned at 60 years.

How did the researchers interpret the results?

The researchers in the modelling study said that the findings highlight several areas of CT scan use that can make large contributions to the total cancer risk. They also said that risk reduction efforts may be necessary for people in certain age groups who receive the largest numbers of scans, and where high radiation doses are used.

The cross-sectional study concludes that radiation doses used in commonly performed CT examinations are higher and more variable than generally thought, which they say highlights the need for greater standardisation across hospitals.

Conclusion

The modelling study provides a detailed estimate of the potential future cancer risks based on current age- and sex-specific CT use in people in the US. The following points should be remembered.

• These figures must be considered as estimates only. They are based on data from a variety of different sources, which may result in inaccuracies, particularly as they use risk estimates from a variety of populations exposed to radiation in different ways (for example Japanese atomic bomb survivors in the BEIR report). In addition, the calculated LARs used in the studies should not be viewed as exact patient risks. Despite these limitations, however, they show the trend and give broad estimates of the extent of the risk from this type of radiation.
• The study calculates the possible development of new cancers, but can say nothing about the expected stage and severity of these cancers or their likely mortality.
• In the cross-sectional study, radiation doses varied considerably between the type of scan and hospital where it was carried out and, as the researchers say, these may not be the standard doses used. The study did not investigate the specific indications for the choice of dose.
• The results cannot be generalised outside the US. Other countries, including the UK, may use CT scans much less frequently or use different radiation levels.

It should be emphasised that the risk to individuals who have had CT scans is likely to be very small. The issue these studies are calling attention to is that when more people are exposed to radiation from CT scans, their collective risk becomes higher. As a result, more cases of cancer can be expected to occur. This is an important area of further investigation, as reducing unnecessary scans has the potential to reduce population risk and cancer numbers.

Clinicians should always weigh the risk of radiation exposure from a scan against its benefits. That is, they should ensure that the scan is necessary and that radiological investigations are only performed when the findings have definite diagnostic and treatment implications.