Cancer induction caused by radiation due to computed tomography: a critical note

Computed tomography (CT) is, together with interventional radiology and PET-CT, one of the three major radiological procedures that deliver the highest dose to the patient (1). Due to its accurate and valuable diagnostic information CT has become a tremendous asset in clinical healthcare. This has caused a considerable increase in the numbers of CT procedures performed. For instance, the average annual increase in the numbers of CT procedures over the years 1998–2007 is about 10% in the UK (2). Furthermore, the increase in the numbers of CT procedures from 3 million in the early 1980s to 67 million in 2006 has been documented for the US (3) and for a European country such as Denmark where a 20-fold increase has been measured from 1979 to 2005 (4). These numbers demonstrate the importance of CT in clinical practice, not least by serial CT examinations for the evaluation of disease progression and treatment monitoring. This explosive growth of CT has urged responsible authorities and researchers to evaluate the risk of carcinogenesis in the population in relation to the radiation dose delivered to the patient. Of special interest is the dose delivered to children as lifetime solid cancer risk estimates for those exposed as children might be factors of 2–3 times higher than the estimates for the general population (5). A single patient undergoing CT may receive an effective dose that varies between about 2 mSv (head) to about 20 mSv (CT-based coronary angiography). While at effective doses above 50 mSv the risk of cancer induction increases linearly with dose, this dose-response relation has not been demonstrated at doses below 50 mSv. In fact, below 50mSv no convincing epidemiological evidence for cancer risk exists and calculations on this risk are based on scientifically questionable, if not invalid, extrapolations of data from higher doses (6). This paper summarizes the data mentioned in a number of articles from recent literature discussing cancer risks due to CT and puts the results of these studies in perspective of current scientific knowledge in the field of radiation protection. For this we follow the lead of the International Committee of Radiation Protection (ICRP) and the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR). In this context we critically re-visit various recent articles in peer-reviewed medical journals. We will emphasize the fact that low-level radiation cancer risk estimates that are derived from a risk factor of 5% per sievert based on collective dose are not valid in the dose ranges associated with a single CT examination. Furthermore, in view of the rising number of patients undergoing CT and the growing public awareness of the detrimental effects of radiation from medical imaging procedures we review the strategies and efforts of various national and international bodies and manufacturers of CT apparatus to lower the radiation dose to the patient.

A brief explanation of dosimetric units used in this paper

The absorbed dose refers to the total energy absorbed by the irradiated tissue. It is measured in joule per kilogram and its unit name is gray (Gy). The organ absorbed dose refers to the amount of energy transferred to the organ after it has been irradiated. In order to throw light on the biological effect of a given radiation the equivalent dose has been introduced. It is defined as the absorbed dose multiplied by a quality factor that weights the radiation specific damage. In case of several radiation types and energies the sum of the individual amounts represent the equivalent dose. In the case of X-rays as used in CT the radiation quality factor is 1, so that the equivalent dose equals the absorbed dose, i.e. has the same numerical value. However, because it takes into account the specificty of the radiations, the meaning is different. Consequently, the unit is different, it is the sievert (Sv).

In order to take into account that different tissues do not have the same sensitivity to radiations, the effective dose is used. A specific weighting factor is applied to each tissue. The equivalent dose of each organ is multiplied by the organ weighting factor and the effective dose is the sum of the weighted equivalent doses of all the organs of the body. Thus the effective dose takes into account both the type of radiation and the type of tissue.

A collective effective dose can be calculated as the sum of all individual effective dose. It represents the global doses received by a population. Its unit is man.Sv. The man.Sv/capita, i.e. the collective effective dose divided by the population of a country, defines the average dose received by a single individual. It is an indicator of the exposure of a population.

The linear no-threshold model

The effect of low amounts of radiation, viz effective doses below 100 mSv in adults and 50 mSv in children, is a subject of continuous debate. Epidemiologic studies from A-bomb survivors demonstrate that the excess risk of cancer is linearly related to the effective dose. Furthermore, this relationship is observed in situations where the exposure to ionizing radiation is delivered at high doses rates.

For the purpose of radiation protection, the international consensus has been to adopt the low-dose range, whatever the dose rate, a linear no-threshold (LNT) model. It is not scientifically established, but it is practical to set up radiation protection rules. The LNT model uses the hypothesis that any radiation is potentially harmful and requires some kind of protection. The excess risk of cancer is evaluated as 5% per Sv effective dose. Although ICRP has stated that this risk estimate should not be applied to evaluate the excess risk of a population from the collective effective dose, especially for individual trivial doses, some authors regularly report an excess of several thousands of cancers due to medical exposures.

It has been argued that the LNT model overestimates the risk on the one hand or underestimates the risk on the other hand (7, 8).

The arguments in favor of an over-estimation of the risk by the LNT model are as follows: (a) there are no significative effects for effective doses below 100 mSv in adults and 50 mSv in children; (b) epidemiologic studies over very large populations living in geographic areas with high natural background fail to demonstrate any effect; (c) continuous exposures to low doses of radiations provide some protective effects (adaptive effects and hormesis); and (d) oncogenesis with the three steps of induction, promotion and progression is obviously not linear at all.

On the opposite, the arguments in favor of an under-estimation of the risk by the LNT model are the following: (a) the bystander effect is the observation of chromosomic lesions in cells which have not been irradiated but are neighbours of irradiated cells – this effect can be transported by the supernatant of irradiated cells demonstrating that soluble clastogenic factors – not yet identified – carry out the bystander effects; (b) genomic instability is the observation of the amplification of chromosomal abnormalities after a few more cell divisions although those cells have not been exposed to further irradiation. Genomic instability contradicts completely the dogma that DNA duplicates exactly at each cell division. Bystander effect and genomic instability are two phenomena demonstrating the complexity of cellular mechanisms and the necessity of future research.

Finally, there are two remaining issues which makes things even more complicated: (a) DNA lesions resulting from ionizing radiations, i.e. damage to the basis, single strand breaks and double strand breaks, are paving the way to oncogenesis. However, there is a large gap between a DNA lesion and a cancer, lesions of many genes being necessary in combination to produce a cancer; (b) radiation hypersensitivity is a status presented by some 5–10% of the population. Because these persons have some deficiency in the molecular signaling of lesions due to radiations and of their DNA repair mechanisms, they present some degree of hypersensitivity to ionizing radiations.

Cancer risk estimates due to CT; data from literature

Screening with CT

Whole-body CT screening for healthy adults has become of increasing interest especially in the private independent radiology clinics. Such screening aims at early detection of a variety of diseases including colon cancer, prostate cancer, lung cancer and breast cancer. Apart from the non-proven medical benefit of whole-body CT screening, the lung and bone-marrow radiation dose are a major concern. It has been suggested that single whole-body CT examination in a healthy 45-year-old adult results in a lifetime attributable excess cancer risk of about 0.08%. If this annual examination would take place at the age of 75 this excess risk of cancer mortality would be 1.9% (9). Against this background it is worthwhile noting that different CT systems produce different doses and therefore different risks. Also the dose is subject to user-dependent factors such as differences in protocols, focus on optimization, training and competence. Interestingly, also a CT screening program for coronary calcifications in asymptomatic men and women at 5-year intervals has been proposed. Using a median chest dose of 2.3 mSv, the estimated excess lifetime cancer risk is about 50 per 100,000 persons (0.05%) with a range up to 200 (0.2%) for men and 300 (0.3%) for women (10). The authors conclude that the radiation-induced cancer risk associated with CT screening is not negligible.

CT-based patient studies

CT for suspected pulmonary embolism or CT-based coronary angiography, delivers a typical dose of 10–20 mSv to the chest of the patient. An estimation revealed that 1 in 270 women and 1 in 600 men who underwent this examination aged 40 years will develop cancer. For a 20-year-old patient the risk is doubled and for a 60-year-old approximately halved. These lifetime effects were estimated in non-cancer patients in the context of remaining life expectancy (11). Another study has suggested that in nearly 1 million non-elderly adults across the US at least 70% of the patients underwent a 3 mSv examination during a 3-year period, whereas a ‘sizable minority’ received moderate and high radiation doses. A generalization of these data suggests that about 4 million Americans receive a cumulative effective dose that exceeds 20 mSv per year. The highest effective doses were registered for CT examination of the coronary arteries, chest, abdomen, and pelvis (12). Sequential CT has been studied in the Aarhus area in Denmark mapping the radiation risk in patients (n = 300) who undergo more than six CT examinations during one year. An accumulated dose of about 21.5 Sv was registered for the total of all 300 patients. The authors estimated that 1 in 265 female patients and 1 in 344 male patients had a lifetime risk of developing solid cancer. When all CT examinations in 2005 were totalled (n = 3769) a mean effective radiation dose of 18.9 mSv per patient was calculated. It was mentioned that this radiation dose will induce a clinically relevant cancer in seven patients (4).

Radiation-induced cancer in children attributed to CT

With a proper medical indication the potential benefits of CT examination in children are clinically recognized and documented provided that the radiation dose is minimized (13, 14). It has been determined that in the US approximately 600,000 abdominal and head CT examinations are performed annually in patients under the age of 15 and it has been estimated that 500 of these patients might ultimately die from CT-induced cancer (15). The dose resulting from CT radiation exposure for children under 15 years is strongly age-dependent. For example, a retrospective investigation in the Hospital for Sick Children in Toronto, Canada estimated the effective dose from head CT in neonatal and 1-, 5-, 10- and 15-year-old age groups. These numbers were 4.2, 3.6, 2.4, 2.0 and 1.4 mSv, respectively. For abdomen/pelvis CT the effective doses were 13.1, 11.1, 8.4, 8.9 and 5,9 mSv, respectively (16). On the basis of these results it has been concluded that the highest risks are present in the youngest age groups. Furthermore, the authors suggest that the estimated lifetime cancer mortality risks due to CT radiation in a 1-year-old amounts to 0.18% (abdominal) and 0.07% (head) (15). In patients with cystic fibrosis who annually undergo repetitive CT from the age of 2 years until death mortality due to CT-induced cancer has been estimated on the basis of a radiation dose to the chest of 1 mSv. For a median survival of 26 and 50 years the survival reduction was 1 month and 2 years, respectively (17).

Examples of fatal cancer risk based on national collective dose due to CT

Preliminary results from the US reveal that in 2006 the per capita dose from medical exposure (excluding dental and radiotherapy) has increased almost 600% since 1982 to about 3.0 mSv. The 62 million CT procedures accounted for 15% of the total number and more than half of the collective dose. This would imply that the collective dose due to CT was about 450,000 person-Sv (18). By ‘simply, but inappropriately’ (see below) applying the risk factor for fatal cancers pointed out by the ICRP recommendation of 1990 amounting to 5% per sievert (19) this collective dose would result in about 22,500 fatal cancers per year.

Cancer risk in the light of ICRP recommendations 103 (2007) and UNSCEAR report

At the international level radiation risks are reviewed regularly by the ICRP and UNSCEAR. Both bodies aim to assess the risk of radiation-induced cancer and base their conclusions on ongoing information obtained from epidemiological studies on radiation-exposed groups. Their estimations on response to radiation dose are largely based on the follow-up of the survivors of the atomic bomb explosions at Hiroshima and Nagasaki, i.e. the so-called Life Span Study (20, 21). Their studies comprise large cohorts and provide reliable data on the excess of carcinogenesis over age in survivors. However, these are not persons who were medically exposed to diagnostic radiation. Also other factors such as dose rate, volume of exposure and exposure duration are different. Both ICRP and UNSCEAR appreciate the fact that cancer risk assessment at low medical doses is hampered by lack of epidemiological data, limitation of statistical precision and the need to perform multiple statistical tests in an attempt to establish a minimum dose at which elevated risks can be detected. Whereas recent data for the survivors of the A-bombings point to a linear or linear-quadratic dose response trend such a relation has not been established in the medical low-dose range below 50 mSv. In particular, it will be virtually impossible to establish whether a small increase in cancer risk exists due to medical exposure because of lack of statistical power as the incidence of cancer in any population is high. In addition, the population that undergoes medical exposure is different with respect to age distribution and health conditions. This also implies that the issue whether there is a threshold for risk cannot easily be solved. Yet ICRP publication 60, published in 1990 (19), provides a method to calculate the number of fatal cancers on the basis of a collective dose, i.e. 5% per sievert. In their publication Hall and Brenner (15) emphasize that this risk estimate is crude and most likely not applicable in low-dose medical diagnostics for two important reasons: (a) it assumes the validity of the linear-no-threshold (LNT) hypothesis at low doses; and (b) it assumes that there is no variation of radiosensitivity with age and gender. We want to accentuate that the LNT is a hypothetical concept and not a scientific finding (22). As explained above, it has been formulated as a way to implement radiation protection rules regarding a population and should not be used to calculate or estimate cancer risks of low-dose diagnostic medical radiation. At present there is evidence that cell and tissue responses to radiation damage, particlularly at low doses and/or dose rates are non-linear and may exhibit thresholds (22). Accordingly, in their publication 103, article 66 the ICRP states:

‘However, the Commission emphasises that whilst the LNT model remains a scientifically plausible element in its practical system of radiological protection, biological/epidemiological information that would unambiguously verify the hypothesis that underpins the model is unlikely to be forthcoming. Because of this uncertainty on health effects at low doses, the Commission judges that it is not appropriate, for the purpose of public health planning, to calculate the hypothetical numbers of cases of cancer or heritable disease that might be associated with the very small radiation doses received by large number of people over long period of time.’

Moreover, article 161 of ICRP publication 103 reads: ‘… Specifically, the computation of cancer deaths based on collective dose involving trivial exposures to large populations is not reasonable and should be avoided…’ (6). Indeed, risk estimates should be based on organ-specific risk factors and age- and gender-specific risk factors.

Strategies for reducing the radiation dose in CT

The failure to demonstrate that a risk of cancer exists does not mean that the risk does not exist. The increase of the exposures of the population due to medical diagnostic radiation will continue and over time it may be possible to demonstrate an excess of cancer by epidemiological studies with sufficient power since individual effective doses may cumulate up to the level of 50–100 mSv. At present, however, this does not relieve us of the task and obligation to slow down the increase of radiation exposure. Therefore various actions should be undertaken and should be based on the guiding principles of radiation protection in medicine :

Correct and justifyable medical indication;

Performance of the examination according to the ALARA (as low as reasonable achievable) principle;

Performance of adequate quality assurance;

Specific training for professionals involved in radiological patient examinations.

As the increase in the population dose is largely caused by CT various additional measures are relevant:

Choose technical parameters according to patient dimensions and anatomic region to be imaged and modulate tube-current-time product;

Consider the right dose for any patient in view of the diagnostic task;

Pay special attential to patient specific parameters as pregnancy and age;

Use technical possibilities as ECG-based tube current modulation for cardiac studies;

Re-visit referral criteria for imaging by highlighting when MRI and ultrasound must be preferred to ionizing radiation;

Produce patient dose in standard DICOM format and record individual patient dose and technical parameters used;

Provide information for patients regarding the risk and net health benefit of diagnostic radiation.

Initiatives with regard to dose reduction

Alarmist public articles on the cancer risk of CT usage have caused a negative perception of imaging and radiation and have caused unnecessary stress to patients who may even decline a properly justified imaging test. One article (3) has stated that 1.5–2.0% of cancers in the US may eventually be caused by ionizing radiation used in CT. The positive effect of this negative media attention has been important efforts to minimize the CT dose. In 2010 the FDA is launching a collaborative ‘Initiative to Reduce Unnecessary Radiation Exposure from Medical Imaging’ with a focus on those procedures that cause the highest radiation doses including CT. This initiative is assisted by groups including the ACR, the AAPM, ACC and the NCRP (National Committee for Radiation Protection). One goal is to develop ‘appropriate use criteria’ linked to a number of medical conditions. The careful use of CT in pediatrics has been addressed by the ‘Image Gently and Step Lightly’ campaigns by the Alliance for Radiation safety in Pediatric Imaging. In addition, the ACR and the RSNA are developing an ‘Image Wisely’ campaign, which is also meant for the population of adult patients. Likewise, the European Commission is working on a new directive laying down Basic Safety Standards aiming at the protection of patients and other individuals working with ionizing radiation to replace the previous directives 96/29 and 97/43 Euratom. The EC has also published the Radiation Protection Report 154 on European Guidance on Estimating Population Doses from Medical X-Ray. Together with the RP report 154, the new directive to be published in a couple years form a new basis for the radiation protection of patients.

European professionals in radiation safety have launched the European Medical ALARA (as low as reasonably achievable) Network (EMAN). An interesting initiative is the planned combined effort of Heads of Radiation Control Authorities in Europe (HERCA) and European Coordination Committee of the Radiological, Electromedical and Healthcare IT Industry (COCIR, representing the radiological healthcare industry in Europe) (in particular CT manufacturers), expressing a shared responsibility on patient dose reduction.

Recent technical developments aiming at CT radiation dose reduction

In view of the awareness of dose reduction of CT in the last decade various manufacturers have developed new technology platforms which create benefits for patients. An extensive review on these developments has been published recently by McCollough et al. (23) and Gunn and Kohr (24). Various dose reduction opportunities will be discussed below.

One result has been the ‘Automatic Exposure Control’ technology, which involves the modulation of the tube current according to patient attenuation. The tube potential and the current used for the examination determine the beam energy and the photon fluence and reduction in tube current is the most practical means of reducing CT radiation dose (25). It has been demonstrated that this technology can result in a dose reduction of 35–60%. The drawback of the method is, however, that a signifcant dose increase has been measured for salivary glands (28–48%), bladder (22–51%) and ovaries (24–70%) in relation to the high density of skull base and pelvis bones (26). This implies that the use of this technology in, at least, children and pregnant women should be reconsidered. For these patients the use of low tube current, tailored pitch and limited anatomical range is a better option to reduce the radiation dose.

Another technical development concerns the ‘Adaptive Dose Shield’ technology. This methodology is based on limiting the over-ranging by the X-ray beam. This is achieved with the prepatient collimator that opens and closes at the beginning and end of each spiral CT acquisition. As a result only the relevant image area is exposed to radiation. This technology received FDA clearance in 2008.

A technique of considerable interest is based upon iterative reconstruction as opposed to the currently used filtered back projection. Although this promising method is still under continuous development, it has already resulted in visually sharp anatomic structures for abdominal (27) and chest (28) examinations. It goes together with a substantial reduction in abdominal dose of 31–64% for adults (29). Iterative reconstruction was first used to improve image quality in nuclear medicine. Scintigrams are typically produced by low photon numbers from patients to whom radiopharmaceuticals have been administered. Iterative reconstruction starts with a reconstructed image from the measured projections. Next a ‘re-projection’ process calculates new projections, simulating a CT measurement on the basis of the original image. The deviation between the original image and the calculated projections serves to reconstruct a corrected image. In a loop-wise mode the updating of the image occurs by moving data back and forth. The result is characterized by enhanced spatial resolution in image areas with higher contrast and by reduced noise in areas with low contrast. Thus, iterative reconstruction is especially useful in low-dose imaging (30). However, at present there are no large patient studies that prove that iterative reconstruction provides sufficient diagnostic quality and yield compared to traditional methodology, although preliminary studies are encouraging (31, 32). This makes it uncertain whether this technique can soon be introduced in everyday patient care.

With regard to cardiac CT an important method is based upon the continuous monitoring of the ECG (‘ECG-triggering’), which is used to adjust the tube current. An algorithm that predicts the ‘phase of interest’ of the patient's cardiac cycle drives the tube current to 100%, whereas during the remaining part of the cycle the current can be reduced to below 10%. The development of versatile ECG-pulsing algorithms makes it possible to use this method as well in patients suffering from arythmia. This ECG-triggered acquisition has shown to reduce the radiation dose by up to 90% for CT angiography. This method uses a typical scanning time of 0.25–0.27 s in fast table movement, which allows the entire heart to be scanned in a fraction of a heartbeat. Compared to retrospective ECG- triggering a dose reduction up to 90% has been assessed (33). Recent clinical results obtained by Alkadhi et al. (34) have demonstrated that 128-slice dual source CT coronary angiography provides high diagnostic accuracy for the assessment of significant coronary stenosis, the high-pitch mode further lowering the radiation dose relative to the step-and-shoot mode. The effective radiation dose with the former technique was 0.9 +/– 0.1 mSv and with the latter technique 1.4 +/– 0.4 mSv whereas high diagnostic image quality was obtained in 98.9% and 98.6% of the images, respectively.

A fundamentally different and encouraging dose reduction system involves ‘320-Row Detector’ technology that prevents overranging and intrinsically offers high resolution imaging. This multidetector technology is especially useful for pediatric CT.

In this context the recent availability of a multidetector CT scanner is of special interest as it allows axial volumetric scanning of a 16 cm long range in a single 0.35 s rotation with an acquisition configuration of 320 × 0.5 mm. As to neonatal and pediatric imaging this allows volume scanning with no over-ranging, reducing the radiation doses considerably (32). An additional advantage is the ultrashort scanning time which reduces motion artifacts and may reduce the need for sedation. It has been determined that with this axial volumetric scanning dose savings vary between 18–55% as compared to helical scanning (35, 36).

Apart from favorable pediatric applications this 320-row multidetector system has recently proven its clinical validity in cardiovascular applications including global ventricular function (37), coronary angiography (38), assessment of coronary artery calcium (39), and assessment of significant coronary artery disease (40). Moreover, for single-heartbeat CT angiography this way of volume scanning offers a reduction of the effective dose up to 91% relative to helical scanning with similar image noise (41).

Conclusion

Radiological imaging results in a net health benefit for patients provided the radiation dose is as low as reasonably achievable. It should be emphasized that the reduction in radiation dose is of special importance for children undergoing CT examination as their lifetime risk for radiation-induced cancer is higher than for adults. Yet, recent publications in the medical literature give rise to unfounded negative publicity. Numbers on cancer risks that are derived from the above mentioned ‘5% per sievert rule’ are not scientifically valid as this method of calculation has never been meant for the low radiation doses that are used in medical diagnostics, i.e. below 50 mSv. More than that, any estimation or calculation on cancer risk related to medical diagnostic radiation is not scientifically founded as no solid epidemiological data exist for these low doses below 50 mSv. However, although the present knowledge about cancer risk due to diagnostic ionizing radiation does not give us a reason to raise the alarm, this may be the opposite in the nearby future due to the ever expending availability and range of medical applications of CT. It cannot take long before epidemiological data at doses around and below 50 mSv is available. In view of the tremendous health benefits of CT all initiatives to lower the dose to the patient are relevant. Many more dose-lowering innovations are expected in the next few years. The most important present development is the recognition of the necessity to prevent CT radiation overdose and the fact that measures have to be taken (42–44). Finally, the effective dose of patients who are repeatedly examined by medical imaging may be well above the values of 100 mSv in adults and 50 mSv in children. Above this level of dosage, which is no longer the low-dose range, uncertainties disappear and epidemiologic studies may well demonstrate in the future an excess of cancer resulting from medical exposures. Thus, medical exposures must be kept under control by the radiologists.

Conflict of interest: None.