PROLARM: Cancer risk from medical diagnostic exposures is strongly dependent upon patients' prognosis

Full text: Purpose: a) To evaluate the impact of the reduced life expectancy of patients (compared to a non-patient group with same age distribution) on their risk of developing cancer from the diagnostic use of radiation. b) To find an approximation to such reduction in risk which depends only on the patient's age, a, and his life expectancy, but is independent of the choice of values for the baseline risk of cancer incidence, m(a), and the enhanced relative risk ERR(a) from radiation exposure. Method: The lifetime attributable risk LAR (of a radiation-induced malignancy to manifest itself) is a function of age at exposure, e, and given by integrating over attained age, a, the product of ERR(a), baseline cancer risk m(a) and the relative probability of surviving to age a, S'(a,e). We define a 'prognosis-based LAR modifier' (PROLARM) as the ratio of risks for non-patient, LAR(a), and patient, LARp (a), a dimensionless quantity that gives a measure of reduction of LAR due to the patient's prognosis. With the survival of the patient group, Sp' (a,e), and for any choice of fitted function for ERR(a) like those used in BEIR VII report, PROLARM ≥∫d'(a,e) da/∫Sp'(a,e) da, i.e. the ratio of the survival integrals gives a lower (thus conservative) estimate of the reduction in risk. Results: The method was applied to n=4285 patients with metastatic breast cancer for whom survival as a function of age at metastasis was known. Figure shows that LAR is decreased significantly for all ages at exposure. At younger ages, this decrease is more pronounced (PROLARM ≥ 20 for e ≤ 65). Example: using ERR values of BEIR VII, the LAR due to 10 mSv effective dose at age a = 50 would drop from 1.2 E-3 for non-patient to 4.3E-5 for a patient, i.e. by a factor (PROLARM) of 29. Using only survival data, that factor is 27 (but no LAR can be computed). In other words: 10 mSv for a patient correspond risk-wise to 0.4 mSv for non-patient. The method can be applied to any pathology when the associated survival is known. Conclusion: Calculations of cancer risk from diagnostic medical exposures must take into account the patient's prognosis. LAR will be significantly lower when patients are severely ill at exposure. (author)