age-based low-dose lung CT?? but more radiation!!

 A recent article in JAMA and referenced in the popular press documented that the majority of lung cancer diagnoses were found in non-smokers, highlighting the limitation of doing low-dose CT scans (LDCT) targeting only smokers; per the US Preventive Services Task Force (USPSTF) guidelines. the researchers suggest a more reasoned approach is to do LDCT more broadly as an age-based approach, as is done for breast and colorectal cancer screenings (see lung cancer aged based CT screening JAMA2025 in dropbox, or doi:10.1001/jamanetworkopen.2025.46222). I will include a second section below summarizing the information we have about the ionizing radiation effects on health problems (this is pretty long, though i do summarize the main points). So, best to get comfy, have a large cup of coffee, and prepare yourself for this blog… (or just skip around). but, those of you enticed to read the whole blog will get a little star by your name...

 

Details:

-- the researchers first retrospectively identified consecutive patients with lung cancer, starting March 20, 2023; patients were selected sequentially back until September 2018 with study follow-up until December 2024. They were then stratified by the 2021 USPSTF low-dose CT (LDCT) screening criteria by applying their screening recommendations (age 50-80 years, ≥20 pack-years of cigarettes, and current smoking or having quit <15 years previously) into "guideline" and "nonguideline" groups (the latter would not qualify per those guidelines). this data was extracted from electronic medical records at a single large academic medical center

-- 997 patients were included: 350 who met guideline USPSTF criteria and 647 who did not (the nonguideline group) 

-- median age 67; 58 women (52% in guideline group and 61% nonguideline), Asian patients 7.5% (3.7% in guideline and 9.6% nonguideline), Black patients 16% (18% in guideline and 15%), Latino 4.1% (5.4% in guideline and 3.4% nonguideline) and 68% White in both groups

-- never smokers 25% (0% in guideline and 38% nonguideline), current smokers 17% (39% in guideline and 4.3% nonguideline) and former smokers 59% (61% in guideline and 57% nonguideline)

-- pack-years of smoking: 30 (40 in guideline and 20 nonguideline groups); quit time of smokers to time of diagnosis of lung cancer: 14 years (3 in guideline and 24 nonguideline groups)

-- pretreatment FEV1.0: mean 80% (70% in guideline and 86% nonguideline)

-- pretreatment DLCO 68% (64% in guideline and 71% nonguideline)

-- previous malignant cancer: 77% (81% in guideline and 74% nonguideline)

-- coronary artery disease:  38% (42% in guideline and 36% nonguideline)

-- lung cancers diagnosed:

    -- adenocarcinoma: 66% overall; 55% in guideline group, 72% nonguideline

    -- squamous cell: 15% overall; 21% in guideline group, 11% nonguideline

    -- small cell: 7% overall; 13% in guideline group, 3.7% nonguideline

    -- non-small cell: 7.3% overall; 8.9% in guideline group, 6.5% nonguideline

    -- large cell: 1.4% overall; 1.1% in guideline group, 1.5% nonguideline

    -- carcinoid: 2.9% overall; 0.6% in guideline group, 4.2% nonguideline

        -- the nonguideline group had more adenocarcinoma diagnoses (469 of 647 patients [72.0%] vs 192 of 350 patients [55.0%]; (adenocarcinomas have a better survival rate than squamous cell carcinomas); and (perhaps related) had better overall survival (median [IQR], 9.5 [6.6-12.3] vs 4.4 [3.7-6.0] years; HR, 0.67; 95% CI, 0.55-0.82; P < .001) than the guideline group.

--​ of ever-smokers (247 of 997 patients [24.8%]), those who quit longer than 15 years previouisly (134 of 997 patients [13.0%]), those with less than 20 pack-years (65 of 997 patients [6.5%]), and those aged outside the 50 to 80 year range (41 of 997 patients [4.1%]) were excluded since they did not meet USPSTF inclusion for smokers

 

Primary outcome: the proportion of patients meeting USPSTF criteria. Secondary outcomes included survival, clinical characteristics, and modeled impact of expanded screening scenarios on lung cancer detection, cost-effectiveness, and risks

 

Results:

-- there were some important baseline differences in the nonguideline vs guideline groups: 

    -- pulmonary function was better: withFEV1: 86% vs 70% and DLCO: 71% vs 64%

    -- prior malignant cancer: 26% vs 19%

    -- type of cancer: more of the less aggressive adenocarcinoma in the nonguideline group (72% vs 55%):

-- of the 997 patients who had lung cancer in this cohort, only 350 (35.1%) met the USPSTF criteria for screening; 647 (64.9%) did not

    -- of note, the nonguideline group included 247 people (38.0%) who stated they were "never-smokers", and those who did smoke, pack-years of smoking reported was much lower than the guidelines group (20 vs 40 pack-years, and longer cessation duration 24 vs 3 years)

-- the types of cancers found differed, comparing nonguideline vs guideline groups:

    -- adenocarcinoma: 72.0% vs 55.0%

    -- squamous cell carcinoma: 21.0% vs 11.0%

    -- small cell carcinoma:  13.0% vs 3.7%

-- cancer stage assessment, comparing nonguideline vs guideline groups:

    -- no significant difference in Stage IV disease

    -- stage I disease (the major curable one): 179 of 647 patients (28%) vs 72 of 350 patients (21%)

 

-- expanding inclusion criteria to age 40 to 85 years, 10 or more pack-years, and no smoking cessation limit, increased detection rate to 62.1% (619 of 997 patients)

-- but a modeled age-based screening (40-85 years) captured 93.9% of cases (936 of 997 patients) and prevented 26,124 deaths annually (20,000 to 32,248 deaths annually), at $101,000 per life saved ($82,000-$120,000)

-- modeling of risks associated with procedures to diagnose benign disease:

    --assuming 14.4% participation, lung screening would generate:

        -- for lung cancer diagnoses: 3.3 million false-positive results annually (2.4-4.3 million results), or 14.7% per person screened (10.7%-19.2% per person screened)

            -- this could lead to 260 000 invasive procedures (150,000-400,000 procedures), or 1.2% per person (0.7%-1.8% per person), with 1000 complications (500-1800 complications), or 0.004% per person (0.002%-0.008% per person)

    -- assuming 70% participation, lung cancer screening would generate:

          -- false-positive results scaled to 16.0 million (11.1-20.9 million results), procedures to 1.26 million (0.73-1.94 million procedures), and complications to 4900 (2400-8700 complications), with unchanged per-person risks

    -- as a reference here: 

        -- breast screening yields 3.3 million false-positive results (2.3-4.3 million results), or 7.5% per person, with 33 000 biopsies (16,000-65,000 biopsies), or 0.08% per person, and 500 complications (200-1000 complications), or 0.001% per person

        -- colorectal screening yields 2.1 million false-positive results (1.5-3.2 million results), or 2.9% per person, with 420,000 polypectomies (270,000-64,000 polypectomies), or 0.58% per person, and 3200 complications (1800-4800 complications), or 0.004% per person

-- radiation risk assessment:

    -- individuals screened every 10 years (they project for 85% of those being screened) from ages 40 to 85 years would receive 5 LDCT scans totaling 6.5 mSv. (an mSv is a millisievert, where a Sievert is a measure of the potential biological health risks)

    -- the 15% requiring annual surveillance would accumulate 26 to 59 mSv. Lifetime radiation-induced cancer risk:

        -- screening starting at age 40: male risk 0.040%, female risk 0.055%, combined risk 0.048%, expected cancers per 100,000 screened: 48

        -- screening starting at age 50: male risk 0.032%, female risk 0.044%, combined risk 0.038%, expected cancers per 100,000 screened: 38

        -- screening starting at age 60: male risk 0.024%, female risk 0.033%, combined risk 0.029%, expected cancers per 100,000 screened: 29

        -- screening starting at age 70: male risk 0.016%, female risk 0.022%, combined risk 0.019%, expected cancers per 100,000 screened: 19

Commentary:

-- the background here is that lung cancer is really common and, in the later stages, is rarely cured, lung cancer being the leading cause of cancer-related mortality world-wide (more than that of breast, prostate, and colorectal cancer). Overall 5-year survival rates remain very low, at 21%, since 75% of patients are diagnosed at advanced stages

-- smoking is only one of many causes of increased lung cancer risk, so it is not so surprising that non-smokers can have lung cancer. it therefore makes sense that in the general population, non-smokers would be a larger lung cancer group than smokers, as found in this study

    -- 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; passive exposure to cigarette smoke, chest radiation, having HIV (even well-controlled) and perhaps HPV, EBV, and Merkel cell polymavirus (https://pmc.ncbi.nlm.nih.gov/articles/PMC7585805/ ); 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. and, of course, smokers who do not qualify to the USPSTF inclusion criteria

        -- radon is the second most common cause of lung cancer in the US behind smoking

-- this current study found a few important things that would support an age-based approach to lung cancer screening:

    -- most lung cancer (64.9%, in the above study) would be missed with the USPSTF recommendations for lung cancer screening

    -- these USPSTF guidelines from 2021 depend on substantial tobacco smoking (screen those aged 50-80 with a ≥20 pack-year history and current smoking or who quit within the last 15 years)

    -- the nonguideline group had higher numbers of Asian patients, women, never-smokers, and long-time former smokers, groups who would therefore not get systematic LDCT for early lung cancer detection

    -- doing LDCT screening in the nonguideline group would therefore find more earlier-stage lung cancer diagnoses and, thereby, should improve survival by the modeling

    -- and the age-based approach would include smokers who minimize their smoking history: the self-reported quantification of lifetime cigarette exposure using pack-years has been found to have a 40% discrepancy between reported and actual smoking history

    -- and this study found that economically it made sense:

        -- age-based lung cancer screening: $101,000 per life saved ($82,000-$120,000)

        -- age-based breast cancer screening: $890,000 ($700,000-$1,100000)

        -- age-based colorectal cancer screening: $920,000 ($700,000-$1,200,000)

            -- though one could argue that mortality is not the only outcome of these cancers: morbidity and disability are very important and taken into account, especially since the morbidity may well vary per the different cancers

    -- age-based screening at a 30% stage 1 detection rate (the study above found a 28% Stage 1 detection rate in the nonguideline group) would be associated with projected saving of $24.76 billion annually in treatment cost, 10-fold the $2.6 billion for the program cost

-- so, this study supports their conclusions that there needs to be more aggressive screening to decrease lung cancer deaths. and this approach should mirror current recommendations for age-based screening as done for breast and colorectal cancer screening

-- as another perspective, only 12.6% of those who should be screened per the USPSTF guidelines are actually screened. it is likely that the percent would be higher if LDCT were normalized for the whole population

-- the initial LDCT screening identified nodules requiring follow-up. Given the clinical improvements in nodule evaluation, there were few biopsies and 92% were managed by more imaging (which, as below, increases radiation exposure a lot and therefore puts patients at significantly higher risk of radiation-induced cancers!!!)

-- there are also incidental clinically important findings in 50% to 60% of those having LDCT. other studies have found coronary artery calcification in 39% thereby likely decreasing cardiovascular events with treatment for this; 24% had emphysema and might benefit from therapy (and, as per below, they are at much higher risk of lung cancer and should be followed closely, even if the emphysema were not related to smoking), 1.9% had abdominal aortic aneurysms, 8-14% had osteoporotic fractures, and 1.7% had other malignancies.. 

-- the initial USPSTF guidelines from 2013 recommended LDCT screening of people age 55 to 80 yo who had a 30 pack-year history of smoking and were either current smokers or who quit within the past 15 years (though these guidelines were updated in 2021 to age 50-80 with a ≥20 pack-year history and current smoking or who quit within the last 15 years, as used in the current study, a notable 5 year increase in potential LDCTs in those continuing to smoke.

-- the recommendations for the inital USPSTF were based on a single 3-year study (the National Lung Screening Trial, NLST), which found a 20% reduction in lung cancer mortality, though there are very real concerns about using a single 3-year study as the basis for the USPSTF guidelines (very unusual for them). for a summary of NLST, see lung cancer CT screen nejm 2011 in dropbox, or 10.1056/nejmoa1102873; the dropbox has a slew of subsequent NLST analyses and other studies in “lung cancer CT” section. NLST in brief:

    -- 53,454 individuals enrolled who were 55-74yo, had at least a 30 pack-years of smoking, and, if former smokers, had quit within the previous 15 years. These individuals were randomized into 3 years of annual LDCT vs chest xray

    -- 24% had a positive LDCT, with 96.4% being false positives

        -- the radiation dose of LDCT was 1.5 mSv, much lower than the 8mSv of a regular chest CT. But given the large number of positive tests that led to many repeat CTs and PET-CT scans, the average radiation dose was actually 8mSv (ie, on average the radiation exposure of full chest CTs)!!

    -- lung cancer incidence was 645 per 100,000-person years who had LDCT, vs 572 cases per 100,000-person years who had chest xray, associated with 247 vs 309 deaths respectively

        -- this translates to a 20% lung cancer death mortality reduction with LDCT, but an absolute decrease of only 62 deaths/100K person-years

    -- for unclear reasons, the USPSTF extended the NLST criteria of people from 55-74yo in NLST to 55-80yo in their 2013 guidelines, creating the potential for 25 years of annual LDCTs in those continuing to smoke, an increase over the initial potential of 19 years in NLST

        -- the NLST did not find a statistically different number of lung cancers found when comparing each individual year of the 3 years of the study (it was 27.3% having a positive result in the first year, then 27.9% in the second, and dropped to 16.8% in the third year). I assume that the basis for extending the LDCT time from 3 years in NLST to potentially 25 years in those continuing to smoke (without any data to support this) was the lack of a significant difference found over the 3 years in NLST (though, in fact, the numbers of positive LDCTs were decreasing, and it would have been very useful to have data of the incidence of positive screens for a few more years than 3...). Of note, the USPSTF did not include any justification for continuing screening for more years, with the potential of 25 years of radiation in those continuing to smoke (and more smoking would also be increasing the likelihood of getting lung cancer, further aggravated by the increased radiation exposure).

        -- as mentioned, it was also not clear why in 2021 they increased the age range further to 50-80yo with at least a 20 pack-year history of smoking  and currently smoking or quit in prior 15 years; this increased the potential number of LDCTs to 30 in those continuing to smoke, an additional 5 years.

            -- and this increase was in the younger age group (who would have more radiation at a younger time, putting them at disproportionately more risk of developing radiation-induced cancers years later, since older people are more likely to die from other causes)

-- so, a few concerns here:

    -- this NLST study led to the projection that there would be one cancer death per 2500 screened by the radiation from the LDCT in just the 3 years of LDCT scans (though no projection to the possible 25 and then 30 LDCTs per the changing USPSTF recommendations): https://pmc.ncbi.nlm.nih.gov/articles/PMC3709596/pdf/nihms465275.pdf 

        -- of note, this number of 2500 is a projection from atomic bomb exposures in Japan leading to cancer and some medical studies (see the section on radiation and health below, which addresses the inadequacies of all of these estimates)

    -- and, importantly, studies have found that some smokers who found out that they had negative LDCTs felt they could just continue smoking…. "well my lungs are fine, so i can continue smoking...." This is especially an issue given how addictive nicotine is and how hard it is to stop smoking.

        -- there is also the added issue that smokers predominantly die from heart disease and not from lung cancer!!! the LDCT guidelines inappropriately bias peoples' perception that the main problem with smoking is lung cancer mortality. this could thereby lead to more cardiovascular disease as people continue smoking

    -- other studies, such as the NELSON, DANTE, and a German study found significant overdiagnosis of lung cancer by LDCT screening, but no difference in all-cause mortality at 10 years (NELSON study), and no increase in lung cancer detection rate after 3-4 years (DANTE study) (https://gmodestmedblogs.blogspot.com/2020/02/lung-cancer-screening-in-smokers.html ), further undercutting the USPSTF's reliance on one 3-year study for their initial recommendations and then grossly increasing the smoking interval

          -- 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 person-years from lung cancer

        -- the USPSTF modified their 2013 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 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. this 2021 guideline was the one used in this age-based study, which also used the probability that the 1 person in 2500 screened would develop lung cancer from radiation. Again, this is not based on any rigorous data and very likely significant understatement. And people with injured lungs from smoking have destroyed cilia (which are important for clearing mucous from the lungs), chronic inflammation (which increases mucous production and also is carcinogenic), COPD/emphysema/chronic bronchitis, more pulmonary infections (which further scar the lungs), loss of alveoli, loss of elastic tissue, pulmonary endothelial dysfunction. And all of this together leads to a 3- to 4-fold increase in lung cancer found in patients with COPD

        -- my review of the justification for the increased screening in the new USPSTF guidelines was that it was based on their mathematical modeling of the benefits vs risks, relying on the flawed assumptions on the radiation exposure. there was no specific justification for increasing the age to 80 (they did quote the results of the NELSON  and DANTE trials, but both enrolled those aged 50-74!!,with apparent findings contradicting the USPSTF findings

    -- 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, when following USPSTF recommendations), 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 unfortunately happened to have been close to an atomic bomb explosion.

         -- and, not pursued in this study, there is the reality that of the 25% of those with abnormal LDCTs (which is a huge number of people if screening is age-determined) need followup CT or PET-CT, the radiation dose for this huge number of patients will increase dramatically and lead to more radiation-induced cancers....

            -- this is a profound issue, since, there would be a whopping 3.3 million false positives by the age-based approach, leading to loads of radiologic evaluations/more radiation, and more psychological distress in those told they had a nodule/potential cancer

    -- also, this current age-based study should be repeated, since it involved only one medical center, and there needs to be confirmation prior to such an ambitious change in health care

    -- there would also need to be mechanisms to decrease implementation barriers (which are an issue for the current recommendations, where many people do not have the financial capabilities for more testing, or do not have accessible facilities to get the LDCT scans (especially in rural, remote areas, awareness of overall preventive guidelines, etc)

    -- and there would need to be mathematical modeling to determine the interval of age-based screenings, with enough information to conclude the risk/benefit of cancer-inducing/cancer-protecting

-- given my concerns on the effects of ionizing radiation, i did a brief summary of a very detailed book chapter on this below. here is a review of their main conclusions; more details after:

 

-- the studies are pretty mixed/flawed. nuclear power plant proximity seems to be less associated with childhood leukemia (considered a "sentinel event" that suggests an association of cancer with radiation exposure). there may be more of an association in those living near nuclear reprocessing plants or if there is radioactive escape from nuclear reactor accidents. the largest and longest database is the survivors of the Hiroshima/Nagasaki atomic bombing in 1945, and this information has been paramount to developing mathematical models of human ionizing radiation toxicity

-- the endpoint of radiation-related deaths is perhaps not the best one (but is the easiest to get), since there may well have been some individuals who may have been treated for cancer and perhaps developed second cancers subsequently or may have had undocumented morbidity/disability from cancer, perhaps beyond the studies' reach

-- and, all of the studies can be criticized, mostly around selection of control groups and documenting their similarity to the exposed individuals, and statisticians are really good at finding problems with all of the studies, since there really are no epidemiological studies that are above reproach (hence, these studies can only confer an association and not causality)

--many studies have been done on childhood leukemia near nuclear reactors in UK, Germany, France, and the US, but with mixed results, since the studies are done in areas with very small populations, and there could be confounders or selection bias related as to why the people living in the proximity to nuclear reactors were living there that might skew the results (perhaps those choosing to live next to a nuclear power plant have less income, more stressful jobs with more industrial or environmental chemical exposures, etc, thereby making it hard to choose a relevant control population to compare outcomes.

-- the best documented effects of radiation are chromosomal changes. per the article referenced below, the authors "concluded that uncertainties due to chance sampling variation in the available epidemiological data are large and more important than potential biases such as those due to differences between various exposed ethnic groups". These authors,  and those in other radiation studies, express reluctance to come to firm conclusions in the non-bomb studies https://www.ncbi.nlm.nih.gov/books/NBK218703/ also, they conclude that "There are a number of important radiobiological problems that must be addressed if radiation risk estimates are to become more useful in meeting societal needs. Assessment of the carcinogenic risks that may be associated with low doses of radiation entails extrapolation from effects observed at doses larger than 0.1 Gy and is based on assumptions about the relevant dose-effect relationships and the underlying mechanisms of carcinogenesis. To reduce the uncertainty in present risk estimation, better understanding of the mechanisms of carcinogenesis is needed. This can be obtained only through appropriate experimental research with laboratory animals and cultured cells", and "In the case of the effects of exposure to low levels of radiation (less than 0.1 Gy, or 100 mSv effective dose), the scientific uncertainty of radiation-induced cancer is considerable as there is little or no empirical knowledge. Despite the uncertainty, decisions are needed for use in setting standards for protection of individuals against the side effects of low-level radiation. Based on current scientific knowledge (or lack thereof), regulatory agencies in the United States currently use a model that describes radiation injury as a linear function of radiation dose that has no threshold; this is called the linear no-threshold (LNT) model. According to LNT, if a dose equal to 1 Gy gives a cancer risk X, the risk from a dose of 0.01 Gy is X/100, the risk from 0.00001 Gy is X/100,000, and so on. Thus, the risk of health effects including cancer risk is not zero regardless of how small the dose is. In the LNT model, data from high levels of exposure where radiogenic cancers have been observed are used to extrapolate risks at lower doses where cancers have not been observed, and if they exist, they are beyond the current science to observe and measure. One result of following the LNT model is that a very small estimated risk, when multiplied by a large number such as the population of the United States, results in an estimate of a substantial number of cases or deaths, which in reality may not exist....Data provided by the updated report of the atomic bombing survivors in Japan continue to be in support of the LNT model across the entire dose range." see below in the part on Japan

     

Radiation/health studies:

-- there was a very extensive review of the many studies done on ionizing radiation and human disease, especially related to cancer: https://www.ncbi.nlm.nih.gov/books/NBK202000/, a pretty long and detailed book chapter on this issue, with a review and critique of the many studies.

-- As noted, though Sieverts and Grays are both used to express the radiation dose quantitatively, they are defined differently: one Sievert is the amount of radiation necessary to produce the same effect on living tissue as one Gray of high-penetration x-rays or gamma-rays, Gray reflects the physical energy absorbed by matter. However, these are functionally equivalent (ie, 1 mSv is functionally equivalent to 1 mGy)

 

Great Britain

-- Sellafield: 14 nuclear installations. initial studies found increased mortality around these installations, but more inclusive later assessments undercut this result, finding that of 60,000 children within 10-km radius, the recorded cancer incidence was 3 cases per year, vs the expected 2 cases/yr. 

-- Dounreay nuclear reprocessing plant in Scotland had a cluster of leukemia;  in Somerset, England and 2 other areas, initial 2 reports found an increased rate of leukemia in children who lived within 16 km radius of the nuclear weapons plant and a nuclear power station. but these increased incidences were not confirmed in follow-up studies

-- a large study of all of England and Wales assessed childhood leukemia and non-Hodgkin's lymphomas and proximity of residence to 23 nuclear installations from 1966-1987, finding no evidence of an increase of either of these childhood cancers

 

Germany

-- Krummel nuclear power plant: 6 cases of childhood leukemia diagnosed, 5 lived within 5-km radius of the plant. however, these cases may be attributable to accidental release of radionuclides from the nuclear research facility nearby, associated with modestly elevated levels of cesium in rainwater and air samples

-- another study of people within 15 km of German nuclear powerplants found an increased risk of all cancers including leukemia, but a further study found no increase

-- there was an excess of childhood leukemia near 2 nuclear reprocessing plants

 

France

-- studies on 4 nuclear power plants and 2 nuclear reprocessing plants did not find excess mortality. one other study found an increased leukemia risk in individuals up to age 25 who lived withing 35 km of a nuclear reprocessing plant, but only in those who frequented local beaches and in those who consumed local fish and shellfish. but, there was concern about control group selection, recall bias, and biological plausibility to be able to confirm causal associations

 

US

--extensive studies performed in 1990 by the National Cancer Institute, the broadest of their kind ever conducted, assessing potential excess cancer deaths in 107 counties containing or closely adjacent to 62 nuclear facilities (52 nuclear power plants, 9 Dept of Energy research and weapons plants and 1 commercial fuel reprocessing plant). all had begun operation before 1982. over 900,000 cancer deaths were found from 1940-1984 vs 3 comparison counties, found negative results, including no excess childhood leukemia incidence

--Cancer risks were also investigated among residents living near the uranium milling and mining operations at Grants, located in Cibola County in New Mexico. Cancer mortality data were analyzed for the period 1950-2004 and cancer incidence data for the period 1982-2004. Lung cancer mortality and incidence were significantly increased among men: standardized mortality rate, SMR = 1.11 (1.02-1.21) and standardized incidence ratio, SIR = 1.40 (1.18-1.64), but not found in women. those living in the three census tracts nearby revealed a higher risk for lung cancer among men, SMR = 1.57 (1.21-1.99). This type of analysis (and many of the above epidemiologic analyses) bring up the potential of ecological fallacy, where population studies do not necessarily reflect the individuals there (eg, a study finding high salt consumption and more hypertension in a community does NOT mean that those individuals eating more salt were the ones who developed hypertension).

    -- However, the excess in lung cancer incidence among men is still likely, since previously reported risks among underground miners seem to provide evidence of increased lung cancer after being exposed to radon and its decay products, coupled with heavy smoking and possibly other risk factors. lots of people were involved in another area; a study was done that included approximately 1,900 men and women who lived in Uravan, Colorado near a uranium mill and nearby mines for at least 6 months within the period 1936-1984 and were alive after 1978. Results showed that among the approximately 450 residents who had worked in underground uranium mines, a significant two-fold increase in lung cancer was found. No significant elevation in lung cancer was seen among the female residents of Uravan or the uranium mill workers. The excess of cancer among uranium miners was attributed to the historically high levels of radon in uranium mines of the Colorado Plateau, and heavy smoking among the workers. A study done in Washington State at the Hanford nuclear site where plutonium was being produced for atomic weapons did not find evidence of a relationship between their calculated radiation dose and thyroid diseases

-- Three-Mile Island nuclear plant accident: the apparently best study on this found an increase in leukemia and lung cancer in areas estimated to have been in the pathway of radioactive plumes

 

Canada

-- a case-controlled study assessed paternal exposure to ionizing radiation with childhood leukemia before the child's conception, finding  a non-statistically significant increase in the leukemia deaths 

 

Spain

-- a study of 7 nuclear power plants and 5 nuclear fuel facilities form 1975-1993 and an updated report from 1975-2003 did not find consistent evidence of increased cancer risk after adjusting for natural radiation and other covariate.

 

Sweden

-- there was a cluster of leukemia  in those <15yo living near 4 nuclear facilities; however there was no consistent evidence for being associated with living in proximity of nuclear power plants

 

Finland

-- no association of leukemia in the vicinity to 2 nuclear power plants

 

Switzerland

-- a well-conducted study of distance of residence at birth from nuclear power plants and cancer, controlled for background ionizing radiation, electromagnetic radiation from power lines etc, carcinogens related to traffic, pesticide exposure, socioeconomic status, population mixing and exposure to childhood infection, found small nonsignficant increases in leukemia, but relatively few patients had the cancer

 

Israel

-- a study near the Dimora nuclear plant in those <25yo who lived within 45km of the station found no excess risk in those near the power plant

 

Japan

-- 16 nuclear power plants were assessed finding no evidence of increased risk compared to control areas among young residents, but leukemia mortality overall was higher among populations living closer to nuclear power plants

-- studies from Hiroshima and Nagasaki after the atomic bomb explosions in 1945, assessed in the life span study (LSS) cohort:

    -- the total number of deaths by the end of 1945 (immediate deaths from the traumatic blast injuries, burns, bone marrow depletion, etc) was around 170,000 people of the total population of 580,000: ie 30% of the overall population!!

    -- 94,000 survivors of the blasts were tracked for mortality, cancer incidence and other health effects

    -- the assessment included survivors who were within 2.5km and those between 3-10km from the blast (the latter, "whose radiation doses were almost negligible"): 86% of the survivors had "colon doses" under 0.2 Sv (74,000 people with 11,000 cancer cases)

    -- the last report was in 2012, covering the period from 1950 to 2003, with findings:

        -- the solid cancer incidence for the period 1958-1998 found an excess of 11% of solid cancers attributable to exposures >0.005 Gy (mean dose 0.23 Gy)

            -- there was a dose-response curve: those receiving at least 1 Gy radiation exposure had a 48% increase

        -- overall, there was a dose-related risk even at relatively low dose levels. For cancer mortality, there was a statistically significant upward curve of mortality with radiation exposure, but this was associated primarily with a sublinear degree of risk in the dose range of about 300-800 mSv and not at low doses.

        -- the highest relative excess was found for bladder, female breast, and lung cancers, followed by cancers of the central nervous system, ovary, thyroid, colon, and esophagus 

        -- the risk of cancer was 50% higher in women; though if not including female cancers, the estimates by sex were comparable

        -- the excess risk was higher in those who were younger at the time of the bomb exposures (more time to develop cancer??) [this also reinforces the concern about the USPSTF lowering the age of LDCT screening above)

        -- both the in-utero and early childhood groups exhibited statistically significant dose-related increases in incidence rates of solid cancer

        -- but, at present, not only is there no evidence to support the hypothesis that in-utero exposure confers greater adult-cancer risk than childhood exposure, and the risk might actually be lower

-- concerns about generalizability of the atomic bomb results:

    -- the radiation exposures were acute, very high levels of radiation that was received in a matter of seconds, and the population was exposed to a small amount of neutrons and not just gamma rays

    -- the fact that the population had to live in a war-torn country where there was malnutrition, poor sanitary conditions, and other severe difficulties makes generalizability of the findings to other populations an issue

    -- these individuals were exposed to whole body radiation, so the results may be different from people with targeted radiation exposure (eg, multiple CT scans of the lungs)

    -- there may be underlying issue muddying the results: those living further away from the bombs may have gone into the higher exposure areas (perhaps to look for relatives) or simply moved to higher or lower exposure areas; the amount of residual radiation (eg, from radiation continuing in the soil) may have affected the results) and make it difficult to translate their findings to lung cancer from CT scans, etc); and some areas had "black rain" (radioactive fallout over time). also studies with such a long followup as these will not include important changes over time, with perhaps people moving out of the area soon vs those staying there and having more continuing exposure, or those getting new occupational or environmental exposures, or those improving their social conditions vs the opposite, or those who started or stopped smoking

    -- the exposed groups in Japan were a general population with unknown predisposing risk factors to cancers. as elucidated below, there are more than additive effects of the results of radiation exposure in those with underlying predisposing conditions (eg underlying lung problems like COPD where radiation increases cancer risk 3-fold or more), so these are different populations and comparisons are fraught.

studies of nuclear workers

    -- studies in the nuclear industry are likely better in defining the effects of low-level exposures over protracted periods of time, vs the LSS data from the atomic bombs

        -- but there is the "healthy worker effect": employed individuals tend to be healthier than the general population, since in the general population there are likely unemployable people because of their underlying illnesses/disabilities

    -- but these studies are actually more accurate in other ways, given the availability of whole-body dosimetry records, the fact that smoking is not permissible during work hours, the long-term employment of many workers, and the accumulation of radiation exposure over time

    -- an Oak Ridge National Laboratory study found a 60% higher rates of leukemia mortality but no dose-response relationship;  one from Rocky Flats had increased mortality from esophageal, stomach, colon and prostate cancers. a study in France found a 40% lower mortality, attributable to a strong healthy-worker effect

Noncancer diseases related to ionizing radiation

-- cardiovascular diseases:

    -- high dose radiation (>30-40 Gy) for treatment of breast cancer and Hodgkin's lymphoma, is associated with a life-long increased risk of fatal cardiovascular events

    -- atomic bomb survivors with lower radiation dose of <5 Gy have found increased risk of cardiovascular disease, with an apparent linear dose response of 14% per Sv, but there was uncertainty with lower doses than that

    -- the data are mixed in non-Japanese epidemiologic studies of lower dose ionizing radiation and cardiovascular disease

-- cataracts:

    -- found in Japan in survivors of high dose radiation, within 3-4 years of exposure

    -- lower doses in Japan and Chernobyl cleanup workers have been associated with cataracts in those who had protracted radiation exposures, though with a threshold of approximately 0.5 Sv

-- thyroid disease:

    -- Chernobyl studies have found a relationship between ¹³¹I thyroid dose and subclinical hypothyroidism, with an excess odds ratio  of 0.10 per Gray, but this was not found in the Hanford Thyroid Disease study

    -- atomic bomb survivors: significant linear relationship for thyroid nodules (malignant and benign), with excess relative risk of 2.01 per Gray

-- hyperparathyroidism:

    -- atomic bomb survivors: increased relative risk of 3.1 at 1 Gy

    -- medical irradiation in a Chicago study of people having radiation exposures to their head and neck areas for benign conditions found excess relative risk of 1.1 at 1 Gy for hyperparathyroidism, but unclear at lower radiation exposures

--neurologic effects:

    -- atomic bomb survivors: high doses of radiation during prenatal exposures were associated with increased risk of decrements in intelligence in kids in those women exposed between 8 and 25 weeks gestation, with diminished school performance and increased risk of seizures

-- life-span shortening:

    -- animal model studies have found small amounts of life-span shortening at high radiation doses

    -- atomic bomb survivors had small decreases in life-span at doses <1 Sv but more at higher doses. 70% of this shortening was from excess cancer risk 

         -- much of the info was from Japan, though some from cluster of cases of childhood leukemia (the "sentinel indicator" as noted above). no evidence for those living in proximity to a nuclear powerplant though there are no direct measures of radiation exposure done. other studies of proximity to nuclear plants and cancers were considered flawed and with mixed results

Accidental radiation exposures to populations

 

Chernoble:

-- the largest accidental release of radionuclides, principally ¹³¹I and ¹³⁷Cs, into the environment in history

-- their findings were population assessments, subject to ecological fallacy (population-based radiation exposures do not translate to the cancer risk of individuals)

-- the major finding was an increase in thyroid cancer, which had a latency of 4-5 years in Belarus, with estimates of at least 2 Gy of  ¹³¹I exposures in several thousand children; this was associated with 4 thyroid cancers per 100,000 individuals per year (vs baseline of <0.05 per 100,000 per year), and this seemed to continue to occur 2 decades after the exposure

    -- though there may have been significant iodine deficiency there, perhaps increasing the likelihood of ¹³¹I uptake and resultant cancers

-- studies have also suggested a 2-fold increased risk of breast cancer in the highly contaminated radiation areas, controlling for improvements in diagnosis and documentation

-- the relationship to leukemia incidence is unclear, especially since there were better screening systems that could account for the increases they found

    -- however, a case-controlled study found that there was a three-fold higher risk in those aged 0-20 if exposed to >10 mSv vs <12mSv

-- the Techa River study in Russia, where radioactive materials were released into the Techa River from 1949-1956, found an excess relative risk of 4.2 (1.2-13 )per Gray for deaths from leukemia

 

Nuclear workers

-- somewhat mixed results, given very different levels of exposures and the "healthy worker effect", reflecting that working people are overall healthier than the general population, and there may have been restrictions on smoking during work hours

  -- a US study of data from 5 nuclear facilities found a 44% increase in leukemia mortality from ionizing radiation exposure

   -- a French study found a statistically significant increase but only in melanoma

    -- increased cancers were also found in Russian workers at Mayak, though not in Chernobyl recovery operations

 

Studies of Medical Exposures to Radiation

-- the amount of medical ionizing radiation exposure in the US has increased dramatically: in 2006, people in the United States were exposed to more than seven times as much ionizing radiation from medical diagnostic procedures than in 1980

-- and radiology begets radiology: many people need followup radiologic procedures because of incidental findings, other types of xrays needed to clarify a finding, more xrays to track progression of an identified lesion, etc

-- there are lots of xrays done that are not indicated, as the medical culture to default to xray has increased. A study of imaging use in the emergency department found high levels of overuse for a variety of reasons: http://ejradiology.com/article/S0720-048X(24)00252-3/fulltext ; another 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

-- an aging population and patients with more chronic diseases adds to the increase in ionizing radiation exposure

-- a dramatic decrease in regulations on industry will undoubtedly lead to more industrial exposures associated with cancer:

    -- there are many occupational exposures associated with lung cancer: uranium miners, exposure to radon, firefighters, coal miners, and pretty much anyone exposed to asbestos, silica, diesel exhaust, heavy metals, several other chemicals (so the list would include those working in shipyards, auto repair, rubber, iron/steel and construction workers, insulators, painters, etc, etc (see https://pmc.ncbi.nlm.nih.gov/articles/PMC3791490/ for a more exhaustive list)

        -- there was a case of severe silicosis recently reported in a worker in Massachusetts in the stone countertop industry: https://www.mass.gov/news/massachusetts-public-health-officials-issue-safety-alert-to-employers-after-states-first-confirmed-silicosis-case-in-stone-countertop-industry

        -- Trump’s embracing “beautiful coal” in his first and second terms, and the associated coal subsidies ($625 million to expand and reinvigorate America’s coal industry”) have been associated with increased black lung disease in Central Appalachia (Kentucky, Virginia, West Virginia)

-- guidelines for radiographic imaging (USPSTF, specialty societies, etc) often suggest imaging

-- though many of the radiation studies as above do have marred methodology and cannot be considered conclusive, we do know:

    -- a recent analysis found that there was a lifetime cancer risk attributable to CT scans in the US that might explain up to 5% of all new cancer diagnoses annually: https://gmodestmedblogs.blogspot.com/2025/05/ct-scan-radiation-increases-cancer.html

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

 

So,

-- the main issue to me is that ionizing radiation toxicity is cumulative (https://pmc.ncbi.nlm.nih.gov/articles/PMC11429844/) and in its totality is increasingly dangerous. which means:

    -- the more radiation exposure quantitatively over time leads to more genetic damage and increases the probability of subsequent cancer

    -- this includes all ionizing radiation exposures over time, not just from LDCTs and the further radiologic testing from a positive result

        -- people are at increased risk if they are getting more regular (higher dose) chest CTs or PET-CT scans because of lung nodules found on LDCTs, huge numbers of CTs, PET-CTs. And 96.4% of positive LDCTs in the NLST study were false positives!!

    -- the LNT model (linear no-threshold, ie there are adverse effects of radiation at very low doses) as proposed above and confirmed in the Japanese studies, further supporting the risk of cumulative ionizing radiation toxicity

    -- cosmic radiation, consisting of high-energy charged xrays and gamma rays, is associated with  annual average exposure being 0.33 mSv: 

https://www.cdc.gov/radiation-health/data-research/facts-stats/cosmic-radiation.html; this is higher in people living at higher altitudes; and there are concerns that cosmic radiation overall is increasing: https://spaceweatherarchive.com/2018/03/05/the-worsening-cosmic-ray-situation/

-- the other important issue is the interaction of risk factors:

    -- there is the multi-hit paradigm for cancer development, that radiation, for example, would have a more deleterious effect in the presence of other risk factors

        -- the array of lung cancer risk factors include: passive exposure to smoke, chronic bronchitis, pneumonia, and tuberculosis; perhaps even taking high doses of B vitamins (B6 and B12); taking beta-carotene supplementation in smokers; chest radiation, having HIV (even well-controlled) and perhaps HPV, EBV, and Merkel cell polyomavirus (https://pmc.ncbi.nlm.nih.gov/articles/PMC7585805/ ); 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

        -- and, it is undoubtedly true that smokers who do not reach the USPSTF LDCT screening criteria (eg smoked a lot but not in the past 15 years) are still at risk for lung cancer, and that risk would be even higher with any of the above risk factors

            -- for example, 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 see https://gmodestmedblogs.blogspot.com/2020/06/copd-in-nonsmokers-inc-risk-of-lung.html

-- as another example, what about all of the mammograms being done?? we have lowered the age of mammograms to 40yo. makes sense, given the increase in breast cancer in younger women. BUT

-- 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... 

    -- 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

        -- and 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 >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

    -- so, are we setting up women to have more cancers in the future because they are getting lots of mammograms, likely many additional diagnostic mammograms because there is such a high rate of dense breasts (especially in those 40yo women who are now being targeted for mammograms that will lead to further diagnostic mammograms), or more radiation because they have large breasts. And now with the addition of more LDCTs as potential lung cancer screenings???. and perhaps for a woman who lives at a high altitude and gets more cosmic radiation exposure? or one who subsequently develops breast cancer and needs mega-doses of radiation therapy later on top of all of their previous radiation exposures????? are they more likely to develop lung cancer in the future???

    -- we need to deal strongly with the fact that 74% of all CT scans that are deemed unnecessary, again adding to the total radiation burden. an important public health issue

-- this all begs the question of why are women getting more breast cancers??? seems likely to me that it is unregulated oncogenic exposures (chemicals in the air, water, ground; perhaps phytoestrogens??). this is a public health issue that needs to be addressed to prevent the cancers in the first place... 

    -- we need much more extensive modeling of a broad assessment of the diversity of potential radiation exposures and subsequent cancers (and not assessing just LDCT screening in isolation), understanding that there may well be very different cancer risks on an individual basis because of their individual risks.      -- this means extrapolating from the totality of potential radiation-induced cancers and not promulgating the likely grossly understated metric of creating one lung cancer per 2500 people getting LDCTs as in this study. it seems to me that we really need a tested, valid risk assessment tool that incorporates this wide array of individual variability and individual exposures in cancer risk

-- one fundamental issue here is that our health care system is biased to early detection and treatment, vs prevention. A functional system, to me, would disproportionately bias a strong public health promotion against smoking (and the vast majority of smokers begin in their teens), with strong limits on accessibility to cigarettes.

         -- and, for some reason, in my experience, many health care insurers now are not covering smoking cessation meds (which really should be free...).  

    --the public health system needs to monitor industrial and other environmental issues that lead to more pollution of air/water/earth.

-- and we should be focusing on reducing general cancer risk factors, since many cancers are associated with diet, exercise, body weight, alcohol consumption, smoking etc.

    -- it should be noted that we in primary care do not have the tools or time to do the type of extensive occupational history needed to assess an individual's risk, though this is clearly an important assessment. Perhaps there could be a validated tool for us to use, and patients could fill out a standardized occupational history form??

  -- so, the real bottom line regarding this current study is that we should be very careful in recommending additional ionizing radiation, given individual susceptibility (eg people with background lung disease or even women with some "benign" breast atypia who have a lower threshold to having cancer with radiation). and we need a public health system that has the will and power and governmental support to identify and correct the many societal issues that are associated with cancer risk

           

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