This paper provides comprehensive estimates of the potential population future cancer risks related to current levels of myocardial perfusion scanning in the U.S. The results suggest that the 9.1 million tests performed each year in the U.S. could result in approximately 7400 [95% UI: 3,300–13,700] additional future cancers. Nearly 70% of these projected cancers were from the most commonly used technetium-99m rest-stress tests and about 30% were from the higher dose dual-isotope studies.
Radiation dose from myocardial perfusion scans varies widely depending primarily on the radiopharmaceutical used, but also on the protocol and administered activity. Previous studies have provided estimates of the radiation doses or frequency of myocardial perfusion scans, but have not estimated the potential cancer risks.3,4,26
The dose estimates from these previous studies are in broad agreement with those presented here. National survey data on the frequency of different radiopharmaceuticals provided key information on one of the sources of variation in dose (). However, as no data are currently available on actual activity levels that are administered in practice, our estimates were based on doses recommended in the ASNC guidelines, which are similar to other guidelines.27,28
The results for cancer risks assumed that the mid-point of the recommended activity range was used. There is anecdotal evidence that the typical administration levels may be nearer the maximum of the recommended values.18
Sensitivity analysis showed that cancer risks would be about 20% higher if the maximum rather than the median dose was used. Conversely, if tests were performed routinely using the minimum recommended activity, radiation exposure could be reduced by about 20% (compared to the mid-point of the recommended range). However, reduction of dose to minimum activity could adversely affect image quality and diagnostic accuracy.29
There are other factors that could reduce radiation exposure from myocardial perfusion scans, such as the properties of the radionuclide itself. For example, use of thallium-201 in dual isotope studies has improved efficiency and throughput in high volume laboratories.30,31
However, thallium-201 has a radiation dose that is typically two-fold higher than technetium-99m because of its longer half-life. Use of thallium-201 has already nearly halved from three million injections in 2002 to 1.7 million injections in 2008;1
probable reasons include the enhanced image quality of technetium-99m and also concerns about the radiation risks. At current radiation doses, Rubidium-82 for cardiac PET has a radiation exposure profile similar to technetium-99m whereas Ammonia -13 has a lower radiation exposure (). Unfortunately Ammonia-13 is not widely used because there is limited availability of cyclotrons necessary for its production. Alternatively, more efficient single photon emission computed tomography (SPECT) cameras, or new generation CT scanners with prospective gating could substantially reduce the radiation dose and hence cancer risk from cardiac imaging.32,33
Although the effective radiation dose from a technetium-99m rest-stress test is slightly lower than that for a typical CT coronary angiogram, the number of cardiac perfusion tests currently performed annually is more than three times higher than the number of CT coronary angiograms.18
Therefore, they make a greater contribution to the collective radiation exposure to the U.S. population and also to the potential future cancer risks from diagnostic cardiac procedures. Using similar methods to those presented here, we recently estimated that these 2.6 million CT coronary angiograms performed in the U.S. in 2007 could result in about 2300 future cancers.5
The comparison of the cancer risks across the different types of cardiac tests by age at exposure highlights the fact that although the effective radiation dose gives a broad indication of cancer risk, it does not take account of the age dependence of radiation-related cancer risks. In particular, because radiation-related breast cancer risk declines for exposures after age 5023
the higher effective dose for a CT coronary angiogram does not necessarily translate into a correspondingly high cancer risk after this age (). Similarly, although the estimated effective dose for a PET scan with Rubidium-82 is slightly higher (15mSv) than for a technetium-99 rest-stress test (12mSv) the risk estimates are lower because the higher effective dose is largely due to the high thyroid dose from Rubidium-82 but in adults the risk of radiation-related thyroid cancer is very small ( and Appendix B
). It should be noted that the dose and risk comparisons for different cardiac tests in were for specific protocols and that exposure levels are likely to vary considerably in practice.
To study the long-term cancer risks from myocardial perfusion scans directly would require a very large sample size (hundreds of thousands of subjects) with long term follow-up.34
Risk projection studies with allowance for the major modeling uncertainties provide a more feasible approach and a more timely assessment of the potential risks. These projections depend on a number of assumptions. A key assumption is the linear no-threshold assumption, which states that radiation-related cancer risks are proportional to dose and that there is no low dose threshold below which there is no cancer risk.35
There is a large body of data to support this assumption, including evidence of significantly increased cancer risks in populations exposed to low-levels of radiation such as nuclear workers and the Japanese atomic bomb survivors.36,37
There is also biological evidence which suggests that it is unlikely that there is a threshold for radiation-related cancer induction.35
Linear risk models fit the available epidemiological data well at these low doses and this model is supported also by experimental evidence.35
As a result most national and international committees support use of the linear no-threshold assumption for radiation protection.8,38,39
However, there is a minority opinion that carcinogenesis has a threshold below which low dose radiation may not be harmful through stimulation of multiple DNA repair mechanisms.40
Because there is evidence that cancer risks from low-dose rate exposures, like nuclear medicine tests, are lower per unit dose than the high dose-rate exposures received by the Japanese atomic bomb survivors, we reduced the risk per unit dose in our calculations by an uncertain factor with a mean estimate of 1.5 (known as a dose and dose rate reduction effectiveness factor).8
Where possible, uncertainties in the calculations were incorporated into the estimates via the use of Monte Carlo simulations.
The life-expectancy of the exposed individuals is one of the key assumptions in these risk projections. The risk estimates in which summarize the risk per 10,000 tests therefore are most appropriate for asymptomatic individuals, i.e. for a group of individuals who will likely have the life-expectancy of the general population. The impact of the assumed life-expectancy on the number of projected cancers from current levels of use () is less straightforward to assess because some of the required data on the life-expectancy of those currently undergoing testing are, by definition, not available. However, we can use a number of sources to address this issue indirectly. For example, we excluded from the calculations the 10% of scans that were estimated to be performed in the sickest individuals, i.e. those who die within five years of undergoing testing. Prognostic studies using myocardial perfusion scans generally report that subjects with normal test results have lower cardiac death rates than the general population (i.e. longer than average life-expectancy), whereas those with abnormal test results have higher cardiac death rates.41
Although there are no nationally representative data on the current proportion of tests that are normal in the U.S., results from a number of surveys in specific settings (e.g. academic medical centers) find that about 40–60% of the patients have normal test results.42–44
Therefore, the impact of the underestimation of projected cancers in those with normal tests and the overestimation in those with abnormal tests may approximately cancel each other out.
Although there was no single data source that included the information required on the current frequency of tests according to age, sex, and test type, the data on age and test type have previously been cross-checked with other sources, including Medicare and the Veterans Association, and showed good concordance.18
As the risk estimates per test were very similar for males and females, it was not necessary to have data on the distribution of tests by sex for our calculations. Similarly, although we did not have data on the number of individuals who underwent tests, only the total number of tests, this will not have affected the estimated potential cancer risks, as at low dose levels the risks are approximately additive. For example, if 4.5 million individuals each underwent two tests the total future projected cancers would be the same as for 9 million individuals who underwent a single test.
Given the multiple indications for cardiac perfusion studies, and the lack of clinical trial data, it has not been possible thus far to estimate the absolute benefits in terms of the number of deaths that may be prevented by these tests.45
However, appropriateness criteria for myocardial perfusion scans have been published by the American College of Cardiology Foundation with support of several organizations.46
In general, perfusion tests were indicated to assess intermediate and high-risk patients with likely coronary artery disease, but they were considered inappropriate or of uncertain appropriateness for low-risk patients or for general screening. A recent multi-center study using these criteria found that about 14% of tests were classified as inappropriate.47
In summary, myocardial perfusion scans are a key tool in the assessment of patients with known or suspected heart disease. For most patients, the risks from not performing the myocardial perfusion scan will be greater than the small radiation-related cancer risks. However, this paper highlights the fact that even when individual risks are small, significant numbers of future cancers can accumulate when large numbers of people are exposed. The estimates depend on a number of assumptions including life-expectancy. They apply directly to asymptomatic individuals with life-expectancies similar to the general population. For individuals with a symptomatic clinical profile, on whom such scans are typically performed, the risks will be lower because of shorter life-expectancy.
The risks could be reduced by decreasing the number of tests performed, for example, by performing stress-only technetium-99m studies, or by decreasing the radiation dose per test. Other modalities that do not involve ionizing radiation such as stress echocardiography or cardiac magnetic resonance imaging (MRI) could be considered dependent on cost, availability, and adequate sensitivity and specificity. Alternatively, newer generation SPECT or CT scanners and hybrid systems, may allow improved detection of disease with lower radiation exposure. For the individual subject, the physician should balance the need for diagnostic testing and the risk-benefit ratio taking account of all potential risks, being mindful of guidelines for radiation safety and appropriateness criteria for the test.