Risk of marrow neoplasms after adjuvant breast cancer therapy: the national comprehensive cancer network experience

Abstract

Purpose: Outcomes for early-stage breast cancer have improved. First-generation adjuvant chemotherapy trials reported a 0.27% 8-year cumulative incidence of myelodysplastic syndrome/acute myelogenous leukemia. Incomplete ascertainment and follow-up may have underestimated subsequent risk of treatment-associated marrow neoplasm (MN).

Patients and methods: We examined the MN frequency in 20,063 patients with stage I to III breast cancer treated at US academic centers between 1998 and 2007. Time-to-event analyses were censored at first date of new cancer event, last contact date, or death and considered competing risks. Cumulative incidence, hazard ratios (HRs), and comparisons with Surveillance, Epidemiology, and End Results estimates were obtained. Marrow cytogenetics data were reviewed.

Results: Fifty patients developed MN (myeloid, n = 42; lymphoid, n = 8) after breast cancer (median follow-up, 5.1 years). Patients who developed MN had similar breast cancer stage distribution, race, and chemotherapy exposure but were older compared with patients who did not develop MN (median age, 59.1 v 53.9 years, respectively; P = .03). Two thirds of patients had complex MN cytogenetics. Risk of MN was significantly increased after surgery plus chemotherapy (HR, 6.8; 95% CI, 1.3 to 36.1) or after all modalities (surgery, chemotherapy, and radiation; HR, 7.6; 95% CI, 1.6 to 35.8), compared with no treatment with chemotherapy. MN rates per 1,000 person-years were 0.16 (surgery), 0.43 (plus radiation), 0.46 (plus chemotherapy), and 0.54 (all three modalities). Cumulative incidence of MN doubled between years 5 and 10 (0.24% to 0.48%); 9% of patients were alive at 10 years.

Conclusion: In this large early-stage breast cancer cohort, MN risk after radiation and/or adjuvant chemotherapy was low but higher than previously described. Risk continued to increase beyond 5 years. Individual risk of MN must be balanced against the absolute survival benefit of adjuvant chemotherapy.

Fig 1.

Patient flow diagram. ALL, acute lymphoblastic leukemia; AML, acute myelogenous leukemia; CLL, chronic lymphocytic leukemia; CML, chronic myelogenous leukemia; MDS, myelodysplastic syndrome; NCCN, National Comprehensive Cancer Network; SLL, small lymphocytic lymphoma.

Fig 2.

(A) Hazard ratios for risk of marrow neoplasm and (B) incidence rates of marrow neoplasm per 1,000 person-years.

Fig 3.

Cumulative incidence of marrow neoplasm. The numbers in parentheses under 5-year incidence and 10-year incidence represent 95% CIs.

Fig A1.

Histogram describing individual times (years) from diagnosis of breast cancer to last follow-up (defined as death, marrow neoplasm [MN] diagnosis, other cancer diagnosis, or last follow-up date at National Comprehensive Cancer Network) of patients in our analytic cohort (N = 20,063). Times were rounded to integer years, so the bars represent values 6 months before and after the given number (eg, patients with times between 1.5 and 2.49 years are included in the bar labeled 2). Most patients (> 12%) have at least 1.5 years of follow-up. However, because the median time from breast cancer to MN was 4.9 years, this longer incubation period suggests that more patients from our study may eventually be diagnosed with an MN, and our current findings may underestimate the true incidence. The latest update on patient follow-up occurred in April 2012 when the database was locked.

Fig A2.

Histogram describing the dates when patients in our analytic cohort presented with breast cancer (N = 20,063).