The risk of cancer attributable to diagnostic medical radiation: Estimation for France in 2015

Abstract

Introduction Although diagnostic medical ionizing radiation (IR) has clear clinical benefits, it is also an established carcinogen. Although the individual IR dose usually delivered from medical diagnostic procedures is small, the number of people undergoing these procedures is large. In France exposure to diagnostic medical IR represented 35% of the French total non-therapeutic IR exposure, being one of the highest exposures to diagnostic medical IR in Europe. This study quantifies the number of new cancer cases in France in 2015 attributable to lifetime IR exposure from medical imaging procedures. Methods The cancer incidence attributable to medical diagnostic IR was estimated using cumulative exposure, risk of cancer for a given dose, and cancer incidence by site. We used national frequencies of medical diagnostic examinations by sex and age reported in 2007 to estimate the lifetime cumulative organ dose exposure adjusted for changes in use of diagnostic procedures over time. The Biological Effects of Ionizing Radiation (BEIR) VII risk models were used to estimate the excess cancer risk due to IR; alongside we used cancer incidence data from the French Cancer Registries Network, FRANCIM. A minimum latency period of 10 years was assumed. Exposures from both external (X-rays, CT scans and interventional radiology) and internal sources (nuclear medicine) were considered. Results Of the 346,000 estimated new cancer cases in adults in France in 2015, 2100 cases (800 among men and 1300 among women) were attributable to diagnostic IR from external sources, representing 0.6% of all new cancer cases (0.4% for men and 0.8% for women). Furthermore, 220 new cancer cases, representing 0.1% of all cases, were attributable to diagnostic IR from internal sources. Overall, the leading cancers attributable to diagnostic medical IR in France were female breast (n = 560 cases), lung cancers (n = 500), colon (n = 290) and bladder (n = 250) cancers. In the sensitivity analyses, we found that the largest differences with the main model were observed when changing the dose-rate effectiveness. Assuming an over- or under-estimation of the dose by 30% also modified the attributable fraction. Using the UNSCEAR model instead of the BEIR VII model increased attributable fraction for all nine considered cancer sites. Conclusion The contribution of diagnostic medical IR to the cancer burden in France is small compared to other risk factors, and the benefits largely outweigh its harms quantified here. However, some of these IR-associated cancer cases may be preventable through dose optimization and more enhanced justification thorough diagnostic examination.