Late subsequent leukemia after childhood cancer: A report from the Childhood Cancer Survivor Study (CCSS)

Background: Subsequent short-latency leukemias are well-described among survivors of childhood cancer. However, late (5-14.9 years from diagnosis, LL) and very late (>= 15 years from diagnosis, VLL) subsequent leukemias have not been well studied. We assessed risk factors, prevalence, and outcomes for LL and VLL in the Childhood Cancer Survivor Study cohort. Methods: Subsequent leukemias, among 25,656 five-year survivors, were self-reported and confirmed by pathology review. Standardized incidence ratios (SIR) and cumulative incidences were calculated, and relative risks (RR) were estimated using Cox regression for exposures. Results: Seventy-seven survivors developed subsequent leukemia, 49 survivors with LL (median time from diagnosis 7.8 years, range 5.0-14.5 years) and 28 with VLL (median time from diagnosis 25.4 years, range 15.9-42.8 years), with a cumulative incidence of 0.23% (95% CI 0.18%-0.30%) 20 years from diagnosis for all subsequent leukemias. The most common leukemia subtypes were acute myeloid leukemia, myelodysplastic syndrome, and chronic myeloid leukemia. Compared to the general population, survivors were at increased risk, for developing LL (SIR 9.3, 95% CI 7.0-12.1) and VLL (SIR 5.9, 95% CI 3.9-8.4). In multivariable relative risk analyses, cumulative epipodophyllotoxin dose >4000 mg/m2 was associated with increased risk for LL and VLL (RR 4.5, 95% CI 2.0-9.9). Conclusions: In this large series of late subsequent leukemias, survivors of childhood cancer are at increased risk, with no evidence of plateau over time. We observed most risk among survivors who received high cumulative doses of epipodophyllotoxins. Ongoing consideration for this late effect should continue beyond 10 years.