Secondary Acute Leukemia in Sarcoma Patients: A Population-Based Study

Purpose: To compare rates of secondary acute leukemia between sarcoma patients and the general population, using data from the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) registry, and to examine whether various patient, tumor, and treatment factors were associated with development of a secondary acute leukemia. Methods and Materials: Patients with a primary diagnosis of connective tissue malignancy between 1973 and 2008 in the SEER database were included. Multivariable competing risk analysis was used to determine risk factors associated with subsequent development of acute leukemia. Using observed-to-expected ratios, we compared incidence rates of secondary acute leukemia between sarcoma patients and the general population. Results: A total of 72,945 patients were identified, with median follow-up of 131 months. On multivariable competing risk analysis, factors associated with increased risk of secondary acute leukemia included receipt of radiation therapy (hazard ratio [HR] 1.67, P=.02), distant disease (HR 2.67, P=.004), male gender (HR 1.53, P=.03), year of diagnosis (HR 0.98, P=.049), and Ewing sarcoma histology (HR 9.95, P <.0001) and osteosarcoma histology (HR 5.06, P=.0001). The observed-to-expected ratio for development of a secondary acute leukemia was 3.67 (95% confidence interval [CI] 1.95-6.28), 3.41 (95% CI 2.73-4.20), and 1.6 (95% CI 1.38-8.19) for acute lymphocytic leukemia, acute myeloid leukemia, and acute monocytic leukemia, respectively. The 10-year cumulative incidence of secondary acute leukemia for patients who did and did receive radiation therapy was 0.3% versus 0.1% (P=.02). Conclusions: Patients treated for sarcoma, in particular those with Ewing sarcoma and osteosarcoma histology, seem to have a higher incidence of secondary acute leukemia as compared with the general population. Treatment factors including radiation therapy and chemotherapy seem to play a role in this increased risk, although the absolute incidence nevertheless remains very small.