Aims To estimate the life attributable risk (LAR) of cancer incidence over a wide range of dose radiation exposure and a large spectrum of possible diagnostic computed tomographic coronary angiography (CTCA) scenarios.
Methods This study included 561 consecutive patients who underwent a successful prospective ECG-gating CTCA protocol (low-dose group) 64-slice CTCA and 188 patients who underwent retrospective ECG-gating CTCA with ECG-triggered dose modulation CTCA (high-dose group). LAR was computed, given the organ equivalent dose, for all cancers in both sexes. LAR was tabulated for each decile of dose-length product by 10-year age classes, separately for each sex.
Results Estimates of LAR of any cancer for an exposure at age ≤40 year were lower in males than in females for any given quantile. At age ≥50years, LAR was similar between sexes only at the lowest exposure doses, whereas at higher dosage, it was, in general, higher for women. At the median age of this case series (62 years) and for a radiation exposure ranging from 1.33 to 3.81 mSv, LAR was 1 in 4329 (or 23.1 per 105 persons exposed) and 1 in 4629 (or 21.6 per 105 persons) in men and women, respectively. For an exposure ranging from 10.34 to 18.97 mSv at the same median age, the LAR of cancer incidence was 1 in 1336 (or 74.8 per 105 persons) in men and doubled (1 in 614 or 162.8 per 105 persons) in women.
Conclusions This study provided an estimate of the LAR of cancer in middle-aged patients of both sexes after a single diagnostic CTCA, providing an easy-to-read table.
- Prospective ECG-gating cardiac computed tomography
- retrospective ECG-gating cardiac computed tomography
- radiation dose
- lifetime attributable risk of cancer
- CT scanning
Statistics from Altmetric.com
- Prospective ECG-gating cardiac computed tomography
- retrospective ECG-gating cardiac computed tomography
- radiation dose
- lifetime attributable risk of cancer
- CT scanning
Radiation dose exposure has been related to an incremental risk of developing a radiation-induced cancer. The life attributable risk (LAR) of cancer incidence and mortality describes an excess of disease cases over a follow-up period with population background rate determined by the experience of unexposed individuals for each of several specific cancer sites at each age of exposure.1 There is a great deal of interest in estimating the LAR of cancer and mortality related to ionising diagnostic examinations.2–10 Computed tomographic coronary angiography (CTCA) has emerged as a noninvasive, patient-friendly diagnostic modality to detect the presence of coronary atherosclerosis11–13; the appropriateness criteria for CTCA have recently been established.14 The potential advantages of CTCA over diagnostic cardiac catheterisation for coronary artery disease assessment have to be carefully weighed against the potential hazards associated with significantly higher radiation exposure during CTCA. Einstein et al15 determined on a theoretical simulation model (using preselected 64-slice scanner parameters and including a reduction by 35% of ECG-gated tube current modulation) the LAR of cancer incidence associated with radiation exposure from a single CTCA study; they showed that the LAR of cancer incidence/mortality rate was considerably greater for women, younger patients and those patients in whom the scan of the aorta was needed. Estimate of the LAR of cancer incidence based upon real scanner data are limited and mostly refer to 16-slice CTCA.16
We aimed to estimate the LAR of cancer using scanner data derived from a large number of patients undergoing clinically indicated CTCA examination and to present them in an easy-to-read chart covering a wide range of dose radiation exposure representing a large spectrum of possible diagnostic CTCA scenarios.
For the purpose of our prospectively designed study, 561 consecutive patients underwent a successful prospective ECG-gating CTCA protocol (low-dose group) 64-slice CTCA at the Fondazione Cardiocentro Ticino, Lugano, Switzerland, because of suspected coronary artery disease or as follow-up after coronary intervention. Moreover, additional 188 consecutive patients underwent a clinically (performed before the introduction of prospective ECG-gating acquisition protocol on the 64-slice scanner) or a technically (inability to reduce mean heart rate or large RR variability) indicated retrospective ECG-gating CTCA with ECG-triggered dose modulation CTCA with the highest tube current output extending from 40% to 80% of the RR interval that resulted in a high radiation dose exposure (high-dose group). Thus, a total of 729 consecutive patients were included in the study.
The standard exclusion criteria for contrast-enhanced CTCA such as previous allergic reaction to iodinated contrast, rapid ventricular response during atrial fibrillation, frequent atrial or ventricular arrhythmias, renal dysfunction (creatinine serum level >1.5 mg/ml), and major contraindications to β-blocker administration (ie, asthma and advanced atrioventricular block) were applied. The institutional review board approved the study protocol.
All the patients received 100 mg of metoprolol (Beloc; AstraZeneca, London, UK) at least 1 h before scanning when needed. If heart rate remained at >65 beats per minute, up to four consecutive intravenous doses of 5 mg of metoprolol were administrated. Heart rate was continuously monitored, and blood pressure was measured at the beginning and at the end of the scanning procedure. In addition, all the patients of the study received sublingual nitrate immediately before scanning. All the examinations were performed with a 64-slice CT scanner (LightSpeed VCT; GE Healthcare, Milwaukee, Wisconsin, USA). The scan sequence included a scout scanogram and a coronary angiogram. A bolus tracking was performed with a region of interest placed into the left atrium, and image acquisition was started 5 seconds after the signal density reached a predefined threshold of 100 Hounsfield units. The calcium scoring was performed in patients >50 years, without coronary stent, with no coronary artery bypass graft (CABG) and in whom no coronary calcium scoring scan was performed within the past 12 months. Lung scanning was also performed in all the patients >50 years old. A total of 70 ml of iodinated, nonionic contrast agent (Optiray 350; Guerber SA, Zurich, Switzerland) was injected continuously into the antecubital vein (80–100 ml at 5.0 ml/s) in patients in whom the native coronary artery was scanned; and a total of 100 ml, in patients with CABG in whom mammary arteries were also investigated. Contrast injection was followed by a 40-ml saline flush injected at a flow rate of 4.0 ml/s by using a dual-headed injector (Optivantage DH; Mallinckrodt, St Louis, Missouri, USA). All the scans were performed in a cranial-to-caudal direction, starting from below the tracheal bifurcation, moving to the diaphragm for coronary and moving down from the sternoclavicular joint when CABG was present.
Retrospective ECG-gating scan parameters (high-dose group)
The scanning parameters used in the high-dose group were the following: tube current was set at 600 mA in patients with a body mass index (BMI) <25 kg/m2 up to 800 mA in patients with a BMI >25 kg/m2 during 40%–80% of the RR interval and from 200 to 400 mA during the remaining RR interval. Tube voltage was set at 100 kV in patients with a BMI <25 kg/m2 and at 120 kV in patients with a BMI >25 kg/m2. Pitch is determined automatically by the system, and it ranged from approximately 0.18 to 0.24, depending on the heart rate.
Prospective ECG-gating scan parameters (low-dose group)
The scanning parameters used in the low-dose group were the following: tube current was set at 600 for a BMI <25 kg/m2 to 800 mA for a BMI >25 kg/m2. Tube voltage was set at 100 kV for patients with a BMI <25 kg/m2 to 120 kV for a BMI >25 kg/m2. The centre of the imaging window was set at 75% of the RR interval for all the participants.
Radiation dose measurements
Radiation dose estimates were expressed by dose-length product (DLP). For each study, the value of DLP was displayed on the console of the CT scanner. The DLP is verified in our institution every 6 months with measurements performed with standard phantoms (BODY 32) and a calibrated pencil camera. DLP was categorised according to the deciles of its distribution (from the 10th to the 100th percentile) over the entire exposed population. The effective dose to individual organs (organ equivalent doses) was calculated in millisieverts for the midpoint value of each quantile. The CT-Expo V.1.6 software (Medizinische Hochschule) implemented with the Monte Carlo method–based program for the conversion factors was used. The software allows sex-specific dose calculation for all age groups and applicability to all existing scanner models including correction of scanner-specific influences. Finally, the total effective dose (weighted equivalent dose) was computed from the organ equivalent doses using the most recent published tissue weighing factors.17
The LAR for cancer incidence was computed, given the organ equivalent dose, for all cancers in both sexes and independently for breast cancer in women according to the methodology described by Einstein et al15 and using table 12D-1 of the National Academies' Biological Effects of Ionising Radiation Seventh Report (BEIR VII) Phase 2.1 LAR was tabulated for each decile of DLP by 10-year age classes, separately for each sex.
The data were summarised as mean (SD) or median and interquartile range if continuous and as counts and percent if categorical. Continuous variables were compared between groups by means of the Student t test; and categorical variables, with the Fisher exact test. The determinants of the absorbed dose at the heart level (heart and calcium score acquisition) were evaluated by means of a multivariable linear regression model. Pearson R (95% CI) was computed to assess the association of the estimate of the LAR (pooled) of cancer incidence and effective dose. The Lin concordance correlation coefficient was used to measure the agreement for measures on a continuous scale. Stata V.9.2 (StataCorp) was used for computation. All tests were 2-sided. A p<0.05 was considered statistically significant.
Baseline clinical characteristics
Table 1 summarises the demographic characteristics of the two groups. Most (90.6%) patients underwent CTCA for investigation of native coronary arteries. Mean age, sex distribution and BMI were similar in both groups. In contrast, mean heart rate was significantly lower (p<0.001) in the patients undergoing prospective ECG-gating CTCA (low-dose group) because of the more frequent use of intravenous β-blockers treatment in this group. The patients with CABG were older than the other patients.
Radiation dose exposure
The mean effective dose in male and female patients of the low-dose group was 3.9±1.7 and 3.1±1.4 mSv (p<0.001), respectively, whereas it was 23.3±4.0 mSv in male patients and 22.3±2.3 mSv in female patients (p<0.023) in the high-dose group. The mean effective dose in the patients with CABG regardless of the scanning protocol was significantly higher (p<0.001) than the dose obtained in the patients in whom only coronary arteries were scanned (table 1). In those patients in whom calcium score assessment was performed (43% of the entire study group), the mean effective doses slightly increased for both prospective and retrospective CTCA (table 1).
Estimation of LAR of cancer incidence
As expected, there was a linear relationship between effective dose and estimate of LAR of cancer with a Pearson R of 72% (95% CI 65 to 77; p<0.001; figure 1). This association was independent of sex, scanning protocol or type of examination (coronary arteries vs coronary arteries in addition to the aorta). On the other hand, the estimate of LAR of cancer incidence, in general, decreased with increasing age at exposure for any cancer and more specifically for breast cancer in women. Tissue-weighted equivalent doses to individual organs from prospective and retrospective CTCA are illustrated in figure 2. The highest weighted equivalent doses were to the lung in male and to the breast in women; these were followed by the bone marrow, the stomach, the thyroid and the liver.
The estimates of LAR of any cancer incidence and for breast cancer incidence are summarised in tables 2 (in male and female patients) and 3, respectively. Estimates of The LAR of any cancer for an exposure at ≤40 years were lower in men than in women for any given quantile, indicating a higher risk for developing cancer in women. At the age of ≥50 years, LAR was similar between sexes only at lower exposure doses, whereas LAR incidence at higher dosage was in general higher for women. At the median age of our case series (62 years old) and for a radiation exposure ranging from 1.33 to 3.81 mSv, LAR was 1 in 4329 (or 23.1 per 105 persons exposed) and 1 in 4629 (or 21.6 per 105) in men and women, respectively. For an exposure ranging from 10.34 to 18.97 mSv at the same median age, the LAR of cancer incidence was 1 in 1336 (or 74.8 per 105 persons exposed) in men and doubled (1 in 614 or 162.8 per 105) in women. The estimate of LAR for breast cancer incidence at the age of 60 years ranged from 1 in 27 777 (or 3.6 per 105 persons exposed), for 1.33–3.81 mSv, to 1 in 2645 (or 37.8 per 105), for 21.39 to 24.93 mSv.
This study estimated, for the first time, LAR for any cancer incidence and for breast cancer incidence using real 64-slice CT scanner readings. We showed that in a typical population (aged between 40 and 70 years) who might undergo 64-slice CT because of suspected coronary artery disease or as follow-up after coronary intervention, the estimate of LAR of any cancer approximately ranges from 1:1500 to 1:4000 for an exposure of 3–5 mSv, which is the most frequent range of exposure of prospective ECG-gating CTCA, but it ranges approximately from 1:300 to 1:1800 for an exposure of 8–25 mSv (which is the most common range of exposure of retrospective ECG-gating CTCA). Tables 2 (in male and female patients) and 3 may be considered a first attempt to provide an easy-to-read chart that can be consulted by any physician to estimate the LAR for any cancer or for breast cancer in a given patient undergoing a single CTCA.
Previous estimates of LAR of cancer were derived either from a theoretical simulation model15 or by applying real data but collected in a limited number of patients and with the past-generation (16-slice) CT scanner.16 The present study confirms previous estimates of LAR of cancer and significantly expands current knowledge. Indeed, our estimates of LAR were obtained in a large patient cohort (the largest one so far) who underwent either a prospective ECG-gating or retrospective ECG-gating 64-slice CTCA and included both patients with previous coronary bypass surgery and patients in whom only native coronary arteries needed to be investigated. This wide range of scanning scenarios makes our data easily applicable to daily practice in which, depending on machine specification and user prespecified protocols, CTCA has been shown to deliver a broad range of radiation doses. As a matter of fact, Maruyama et al18 reported a peak radiation dose exposure using retrospective ECG-gating protocols as high as 43.5 mSv; conversely, CTCA performed with prospective ECG-gating protocol radiation dose exposures may be carried out with a radiation dose exposure as low as 0.8 mSv.19
According to the BEIR VII model,1 estimates of LAR of cancer incidence for women are usually greater than that for men. This is because of the fact that radiosensitivity is much higher for women than for men and that two target organs (ie, lung and breast) are simultaneously exposed in women during CTCA. In their theoretical simulation models, Einstein et al15 suggested that primary contributors to risk at all ages in women were lung and breast cancers, accounting for 80%–85% of all cancers. Our data confirm the high risk of LAR of cancer for women at any age for any exposure with the exception of women aged ≥50 years and for the lowest radiation exposure (1.33–3.81 mSv) where the LAR of any cancer incidence was similar to that of men. These data are novel and might help in reassuring women aged ≥50 years undergoing CTCA that their LAR of cancer incidence is similar to that of male patients provided a low-dose strategy is used. Moreover, it might reinforce the attitude to adopt a radiation-sparing strategy in women to minimise the risk and eventually to equalise the risk of cancer between sex. Our data clearly shows a substantial reduction of LAR for any cancer in the aged population compared with that in the young population, which is easily explained by reduced biological cellular turnover and shorter life expectancy in the aged population. This observation equally applied to women and men.
Awareness of radiation dose exposure among physicians is generally poor as indicated by a recent UK survey.20 The results of the survey showed that most physicians have little knowledge, if any, of the radiation dose and of the potential risk of cancer related to radiological investigations. Moreover, a study from Yale University21 showed that only a limited number of physicians and radiologists could accurately estimate the effect of radiation dose after a CT examination; about 47% of radiologists and only 9% of physicians reported that there was an increased risk of cancer associated with a computed tomographic examination. Altogether, these data suggest that current appreciation of possible radiation-related risk is modest and information about potential risk of cancer is not sufficiently disseminated among the medical community in general and, more importantly, among those physicians and radiologists who either prescribe or are routinely performing radiological examination.22 Among other reasons, there is an intrinsic difficulty in the quantification of the estimate of LAR of cancer incidence (ie, expected number of cases after a single CTCA) and there is no easy-to-read table that can guide the physician. In this view, we attempted to create a synopsis of the estimate of LAR for any cancer for both sexes and for breast cancer in women undergoing a CTCA.
To the best of our knowledge, there are limited data about the awareness of general cardiologists concerning potential cancer risk related to diagnostic procedures using ionising sources, pointing out that 95% of physicians wrongly estimated the risk of fatal cancer with a stress myocardial perfusion scintigraphy procedure.23 Because CTCA may be prescribed also by cardiologists and the risk chart may be equally applied to coronary angiography, we believe that our results may be of interest to cardiologists in weighing potential risk related to single CTCA to other non-invasive diagnostic investigations or diagnostic cardiac catheterisation.
In the target populations for CTCA (ie, those individuals between 40 and 60 years with atypical chest pain and intermediate pretest probability of coronary artery disease and/or those with uninterpretable or equivocal stress test), the LAR of cancer is already very low, ranging from 1:3300 to 1:4329 exposed persons. Thus, these patients are not likely to die of cancer because of a single CTCA. The risk of missing potential relevant information about the status of coronary artery disease is probably greater than the risk of cancer incidence due to radiation exposure, favouring the use of the technique.
Moreover, the newest generation of dual source scanner, which has a greater number of detectors and a faster gantry rotation, can achieve gapless Z-sampling even with a pitch up to 3.2. Using this technology, radiation dose exposure is consistently reduced below 1 mSv in non-obese patients with low and stable heart rate.24 At this level of radiation exposure for a single CTCA, the LAR of cancer for 50-year-old patients (of both sexes) is virtually close to zero as (figure 3). On the other hand, the expected relatively low incidence of cancer demonstrated in our study should not encourage liberal use of CTCA. The Euratom law prescribes that the need for a radiating test should be justified before a patient is referred to a radiologist or nuclear medicine physician; more importantly, any attempt should be made to use a non-ionising technique that may a provide physician with diagnostic information comparable with an ionising investigation.25 It should be emphasised that in the present study, estimates of LAR referred to a single CTCA scan; in contrast, in daily practice, a patient may have had multiple radiating diagnostic tests during workup; thus, the risk of cancer development, as a consequence of multiple exposure to radiating sources, are additive1 and should be always cautiously considered.
Limitations of the study
LAR of cancer estimates using BEIR VII tables are subject to several sources of uncertainty due to inherent limitations in epidemiological data and in the general understanding of how radiation exposure increases the risk of cancer.1 26 In addition to the statistical uncertainty, the populations and exposures from which risk estimates are needed nearly always differ from those for whom reliable epidemiological data are available (ie, survivors of the atomic bombings in Hiroshima and Nagasaki). Moreover, the linear no-threshold models applied in our estimates have also been questioned.27 Despite these uncertainties, the BEIR VII phase 2 model used in this study and in the study of Einstein et al15 remains the most comprehensive and updated assessment of individual risk of cancer for chest irradiation after single CTCA. On the other hand, the present scientific knowledge about risk of cancer development related to exposure to ionising radiation is in favour of the linear non-threshold model.28
We did not compare the CTCA accuracy between patients with low and high radiation exposure. However, in a limited group of patients, a comparison of both prospective and retrospective CTCA data and the intravascular ultrasound (IVUS) data was performed, showing similar accuracy for quantitatve measurements (see Appendix).
Our study provided an estimate of the LAR of cancer incidence based on real data obtained in both sexes after diagnostic CTCA. Moreover, this study provided an easy-to-read table that may help a physician in weighing the potential risk related to a single CTCA with other non-invasive diagnostic investigations or diagnostic cardiac catheterisation.
From our database, 57 patients who underwent both CTCA and IVUS (3.5F, 20 MHz electronic transducer, Eagle Eye Gold; Volcano, Rancho Cordova, California, USA) were reviewed. Of these, 26 underwent prospective CTCA and 31 underwent retrospective CTCA. Quantitative assessment of the lumen area stenosis was performed and defined as the difference between the reference lumen cross-sectional area and the minimum lumen cross-sectional area, divided by the reference lumen cross-sectional area. Lumen area stenosis was calculated in the same manner by two investigators blinded to the type of CTCA acquisition. To ensure that the same corresponding coronary section was taken, side branch or characteristic calcification was used as a reference point. The accuracy of both the prospective and retrospective CTCA data compared with the IVUS data was high and of comparable size: the Lin concordance correlation coefficients for the agreement with the lumen area stenosis assessed with IVUS were 0.82 (95% CI 0.69 to 0.95) and 0.91 (95% CI 0.85 to 0.97).
Competing interests None.
Ethics approval This study was conducted with the approval of the institutional review board.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.