Purpose: To quantify long-term temporal trends in the excess absolute risk (EAR) of secondary leukemia among breast cancer (BC) survivors, using multivariate analyses to evaluate the effects of subtype, age at BC diagnosis, attained age, and calendar year.
Patients and methods: We identified 376,825 1-year survivors of BC within 4 nationwide, population-based cancer registries in Sweden, Denmark, Finland, and Norway (1943-2001). Estimates of EAR (per 100,000 person-years) were modeled using Poisson regression methods and cumulative risks calculated using a competing risk model.
Results: A total of 687 non-chronic lymphocytic leukemias (EAR = 9.05; 95% confidence interval (CI) = 7.5-10.7) was reported. Significantly elevated risks were observed for the first time for chronic myeloid leukemia (CML) (EAR = 2.06; 95% CI = 1.3-2.9) and acute lymphoblastic leukemia (ALL) (EAR = 0.62; 95% CI = 0.2-1.1), in addition to acute myeloid leukemia (AML) (EAR = 5.00; 95% CI = 3.9-6.2). Excesses of CML, ALL, AML and all leukemias combined persisted over 25 years after BC diagnosis. For all leukemias, EAR decreased with increasing calendar year (P = 0.04) of BC diagnosis. Risk for all leukemia and AML by calendar year of BC diagnosis depended on age at diagnosis. For women diagnosed with BC after 1985, the 10-year cumulative risk of leukemia for those diagnosed before and after age 50 was small, 0.10% and 0.14%, respectively.
Conclusions: Although secondary leukemia is a rare event, BC survivors experience statistically significant excesses for at least 25 years after diagnosis, including CML and ALL. Decreasing leukemia risks in recent calendar years likely reflect changes in treatment.