CT scan radiation increases cancer

 a recent analysis found that a lifetime cancer risk attributable to CT scans could encompass up to 5% of all new cancer diagnoses annually (see radiation exposure CT death rate JAMAintmed2025 in dropbox, or doi:10.1001/jamainternmed.2025.0505). thanks to ian huntington for bringing this to my attention

 

Details:

-- about 93 million CT scans (computed tomography) are performed on 62 million patients annually in the US, and these scans are associated with carcinogenic ionizing radiation

-- the assessment below is based on patient-level data from the University of California San Francisco International CT Dose Registry, a conglomerate of CT scans from 143 US hospitals and outpatient facilities associated with 22 healthcare organizations in 20 states

-- each exam was captured in the Digital Imaging and Communications in Medicine (DICOM) metadata which included patient’s age, sex, effective diameter of the body part imaged, scanner type, exam name and description, and technical data such as kilovoltage, milliamperage, scan length, pitch, phase, and collimation

    --the DICOM had the data to develop 26 CT categories reflecting a combination of body regions and clinical indications, with some people having single body regions (e.g. the cervical spine) while other regions had variable radiation doses used depending on the underlying conditions (e.g. for the abdomen, low-dose images included kidneys for kidney stones, moderate doses for trauma, and high doses for cancer)

-- CT utilization was derived from the IMV Medical Information Division CT Market Outlook Report, based on a national annual survey of 235 hospitals and 78 imaging facilities, that was used to quantify the number of CT exams performed in the US in 2023 (this data has been validated against the Medicare and the VA databases)

-- for pediatrics exams in 2022, they accessed the American College of Radiology (ACR) National Radiology Data Registry

-- lifetime radiation-induced cancer incidence was estimated by age, sex, and CT category using the National Cancer Institute software based on the National Research Council's Biological Effects of Ionizing Radiation (BEIR) VII models and projected onto the US population

-- this study excluded CT scans done in the last year of life, which includes 9.6% of all of the CT scans done; they also did not include some of the higher radiation CT scans done for biopsies or other procedures as well as high dose PET-CT scans 

-- main outcomes: lifetime projections of CT-related cancer by age, sex, and CT type/exposure, in male and female adults and in children

 

Results:

-- an estimated 61,510,000 patients underwent 93 million CT exams in 2022

    --2,570,000 (4.2%) were children receiving 3,069,000 CTs

    -- 58,940,000 (95.8%) were adults receiving 89,931,000 CTs

        -- females: 32,600,000 (53%)

        -- males: 28,910,000 (47%)

-- patients received a mean of 1.5 examinations each, with an average of 1 exam per patient across all age groups (though more done in adults vs kids)

-- the numbers CT scans peaked in adults aged 60 to 69

 

-- the doses of radiation used was significantly higher by sex and age:

    -- for example, in children 5 to 9yo, boys had a higher dose of radiation for head CTs (48.0 mGy versus 45.7 mGy for girls); and there was a 29% increase in pancreas dose between boys and girls age 5 to 9

        -- 1 mGy is functionally equivalent to 1mSv

    -- and the mean dose of radiation for routine abdomen and pelvis CT was twice as high in women aged 50 to 60 than in girls age 5 to 9, and organ doses are much higher in children less than one years old

 

-- projected radiation-induced cancers:

    -- overall 102,700 (96,400-109,500)

        -- adults: 93,000 (86,900-99,600)

        -- children: 9700 (8100-11,600)

            -- of note, the absolute risk in children is significantly less than adults, however the relative risks for cancer were higher in children and adolescents

 

-- cases of cancers by type, in descending order of frequency:

    -- lung cancer: 22,400 cases (20,200-25,000)

    -- colon cancer: 8700 (7800-9700)

    -- leukemia: 7900 (6700-9500)

    -- bladder cancer: 7100 (6000-8500)

    -- stomach cancer: 7100 (5500-9100)

    -- thyroid cancer 7000 (5400-9200)

    -- breast cancer in women: 5700 (5000-6500)

    -- liver cancer: 4100 (3400-5000)

    -- kidney, pancreas, oral cavity or pharynx, brain/CNS, esophagus, prostate, ovary, rectum, uterine, were all from 550 to 3000 projected cases

        -- Of note, female breast cancer was the second most common cancer in women, following lung cancer

 

-- Cancers by type of CT scan:

    -- abdominal and pelvis CTs: 32% of all CT scans: 37,500 of the 103,000 cancers (37%)

    -- chest CTs: 21% of all CT scans: 21,500 cancers (21%)

 

 

 

Commentary:

-- of course, CT exams are extremely important in diagnosing potentially treatable conditions, and they will continue to be very important in clinical care

-- however, CT exams do use ionizing radiation, which is clearly associated with increased cancer risk

-- the number of CT exams done in the United States has increased dramatically over the last few decades: the authors note that there were 29,000 future cancers resulting from CT exposures in the US in 2007 and that the number of CT scans performed annually in the US has increased more than 30% since then for a variety of reasons, including larger population, older population, and increased use of unnecessary CT scans

    -- a recent study found that “more than 74% of all CT scans performed on patients proved to be unnecessary”: https://onlinelibrary.wiley.com/doi/10.1155/2023/3709015

-- this study found that, among 62 million people who had CTs in the US in 2023, there were about the 103,000 future cancers that could result from these CT scans 

-- if the number of new cancer diagnoses remain stable (1.95 million in 2023) and the utilization and radiation doses of CTs did not change, then CT imaging could be responsible for 5% of all cancers diagnosed each year

-- as a reference, this is not so dissimilar from the rate of cancers associated with alcohol consumption (5.4% of cancers) and excess body weight (7.6%)

-- given these increases in diagnostic ionizing radiation use, perhaps some of the current findings of earlier age of diagnosis of colon and breast cancers are related to this radiation exposure??

    -- abdominal and pelvis CTs were identified as the CTs associate with the greatest number of cancers, likely being associated with colon cancer (the second most cancer overall)

    -- and perhaps the high use of chest CTs along with lowering age for mammography and increased mammography utilization over time are part of the younger age of breast cancer in women (the second most common cancer in women): https://progressreport.cancer.gov/detection/breast_cancer; the radiation from mammography has decreased with newer machines, but not sure how many of the older machines are still being used; also there are data suggesting that there is more diagnostic mammography being done now: https://www.globenewswire.com/news-release/2025/01/16/3010562/28124/en/U-S-Breast-Cancer-Screening-Diagnostic-Market-Analysis-and-Forecast-2024-2030-Multi-Modality-Imaging-Revolutionizes-Breast-Cancer-Diagnostics-Boosting-Detection-Accuracy.html)

 

-- for adults, the assessment of attributable radiation risk to individuals is largely based on the Japanese atomic bomb survivors, evaluating the rates of different cancers in people at different distances (and therefore quantitative radiation exposures) from the bomb epicenters; the pediatric cancer risk from CTs is available from observational studies)

-- however, i have very specific concerns about using the data from Japanese atomic bomb explosions and assuming that these numbers would apply to CT-induced radiation exposures in patients:

 

    -- increased lung cancer risk in smokers:

        -- smokers who have COPD have a three-fold higher risk of lung cancer versus those without COPD, controlling for the quantity of cigarettes smoked. This likely reflects the fact that both COPD and lung cancer have commonalities, such as both being driven by oxidative stress; both being associated with chronic inflammation; both being linked by cellular aging, senescence, and telomere shortening; both being linked to genetic predisposition; and both having altered epigenetic regulation of gene expression: https://pmc.ncbi.nlm.nih.gov/articles/PMC4718929/ ; also https://gmodestmedblogs.blogspot.com/2020/06/copd-in-nonsmokers-inc-risk-of-lung.html

        -- And other lung cancer risk factors beyond smoking are also elevated in: those with chronic bronchitis, pneumonia, and tuberculosis; perhaps taking B vitamins (B6 and B12); taking beta-carotene supplementation in smokers; having HIV (even well-controlled) and perhaps HPV; exposure to air pollution; occupational and non-occupational exposures, including asbestos, radon, silica, diesel exhaust, burning of wood or coal, and undoubtedly many other inhaled occupational chemicals.

         -- also of note, the NSLT (National Lung Screening Trial, the one study singled out to create the USPSTF recommendations for low-dose CT scan (LDCT) in smokers), they recommended LDCT screening of people age 55 to 80yo who had a 30 pack -year history of smoking and were either current smokers or who quit within the past 15 years.

            -- this NSLT study led to the projection that there would be one cancer death per 2500 screened in just 3 years of LDCT scans (vs the 25 years of the initial USPSTF recommendations)

            -- and, though the LDCTs delivered 1.5 mSv (1 mSv is functionally equivalent to 1 mGy, the measure used in the article above) of ionizing radiation, there was an actual average of 8 mSv (which is the standard dose for a regular chest CT) because of follow-up CTs or PET-CTs for abnormal LDCTs; and there was a very high number of scans given that more than 95% of the abnormalities were false positives and these follow-up studies were normal. future studies did find algorithms that decreased the number of false positives (eg: https://gmodestmedblogs.blogspot.com/2020/02/lung-cancer-screening-in-smokers.html )

          -- however, in their 3-year NSLT study, there was a 20% decreased risk of mortality in those who had LDCTs done, but this translated to an absolute difference of only 62 deaths per 100,000 from lung cancer

    -- there is also the added issue that smokers predominantly die from heart disease. the LDCT guidelines inappropriately biased peoples' perception that the main problem with smoking is lung cancer mortality, and the studies assessing peoples' responses to LDCT results include comments such as "well my lungs are fine, so i can continue smoking...."

        -- the USPSTF soon modified their recommendations to annual LDCT scans for all adults aged 50 to 80 who are either current smokers or quit within the past 15 years. This new recommendation would include many more people and would translate to a potential of 30 LDCTs for those who continue smoking, as well as whatever follow-up high dosage CTs or PET-CTs are necessary, with the attendant risk of very large doses of ionizing radiation over the life of these individuals

        -- and the most recent USPSTF recommendations have lowered the smoking exposure to 20 pack-years, further increasing the numbers of patients getting LDCTs....

    -- the point here is that there are increasing numbers of chest CT exams, and the people getting them are already at higher risk of lung cancer (they are smokers, in the case of LDCT scans), and that radiation on top of abnormal lung parenchyma, ongoing inflammation, etc, predisposes these people to higher risk of lung cancer than the average Japanese person who happens to have been a certain distance from an atomic bomb explosion.

 

    -- increased breast cancer in women:

        -- women with dense breast tissue have a 1.7- to 4- fold increased risk of breast cancer https://academic.oup.com/aje/article-abstract/194/2/441/7726839?redirectedFrom=fulltext&login=false

            -- dense breasts are found radiologically in over 50% in women aged 40 to 50, decreasing to about 40% in women 50 to 60yo, 30% of women 60 to 70yo, and about 25% of women through age 85: https://pmc.ncbi.nlm.nih.gov/articles/PMC4200066/

           -- other known breast cancer risk factors include: air pollution, BMI greater than 30, being tall, having histologic breast atypia, and having a higher BMD (likely reflecting a higher cumulative estrogen exposure)

               -- and, by the way, women with large breasts require a higher radiation dose

           -- ionizing radiation adds to the DNA damage, oxidative stress, inflammation, genomic instability, and interaction with hormonal regulation of the breast: https://www.sciencedirect.com/science/article/pii/S2773160X23000028

-- the issue in these two cases of lung and breast cancer also refer to other cancers: the multi-hit theory for cancer development, whereby adding the oncologic effects of ionizing radiation to those with underlying oncologic risk factors as an additional cellular hit puts people at a much higher cancer risk than those in the general population 

    -- these two examples involved applying the cancer results after the atomic bomb explosions in Japan 80 years ago where the calculus involved the aggregate of presumably lower risk Japanese individuals who just happen to be living certain distances from the epicenter of the explosion. it is a real stretch to apply these numbers to smokers or women (in the above 2 cancers) who are at higher-than-normal risk, live in very different situations and in very different countries with different cultures, have very different exposures, etc.

    -- there is also a small leap-of-faith here since the primary radiation from atomic bombs is gamma rays, whereas the dominant one in CT scans is low-energy x-rays. the general thought is that gamma rays are higher energy, and likely can penetrate tissues more deeply and potentially cause more damage. but it is not clear in the literature which is actually more oncogenic. see https://www.ncbi.nlm.nih.gov/books/NBK304365/ for more on radiation exposure types

-- and, there is also the concern as to how many of our screening interventions actually do lead to lifetime gains (a more striking issue when an ionizing radiation is the screening test), finding essentially none of the screening that we’re actually doing seem to increase longevity: https://gmodestmedblogs.blogspot.com/2023/10/cancer-screening-test-are-they-effective.html

 

Limitations:

 -- several of the limitations and concerns are mentioned above:

    -- the use of atomic bomb data from 80 years ago as our basis for radiation exposure now

    -- the fact that the radiation from atomic bombs (gamma rays) are different from the ones in CT scans (x-rays)

    -- the fact that patients getting CT scans are likely sicker and more susceptible to the cancer risk of ionizing radiation than those who do not:

        -- these sicker patients already have a shorter life expectancy, which might overestimate cancer risk in the future

        -- in particular, these patients often have other oncogenic issues (cellular changes, inflammation, etc) that not only predispose them to cancer, but also increases their chances of getting cancer by the additional oncogenic exposure from CT scans

-- the use of the registries above for data is potentially fraught by inaccuracies that could distort the results

 

So,

-- this study reinforces that we really should limit the number of CT scans we do (ultrasound and MRI are better on this front)

-- which supports the benefits of biennial mammography screening for women, as well as perhaps starting at a later age??? the new USPSTF recommendations suggest starting at age 40 with biennial screening, with only passing comments on radiation exposure. there really should be mathematical modeling of this significant increase in ionizing radiation, as has been done in https://pmc.ncbi.nlm.nih.gov/articles/PMC4878445/ . this really should be part of the argument for or against increasing the number of mammograms done

-- as per above, there is the real issue that these estimates of radiation-induced cancer risk may well be understated, given that the underlying tissues may be compromised by other oncogenic factors (smoking, dense breasts, DNA disruption, chronic inflammation, etc) and the radiation hit may be more oncogenic in patients with already-distorted tissue.

-- this also raises questions about the risk versus benefit calculation for newer tests that are gaining popularity but require ionizing radiation, such as coronary artery calcium scoring, or coronary CT angiography

 

geoff

-----------------------------------

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@bidmc.harvard.edu

to get access to all of the blogs:  go to http://gmodestmedblogs.blogspot.com/ to see the blogs in reverse chronological order

or you can just click on the magnifying glass on top right, then type in a name in the search box and get all the blogs with that name in them