Currently, any myelodysplastic syndrome (MDS) or acute myeloid leukemia (AML) developed after previous exposure to chemotherapy and/or radiation (CT/RT) are defined as therapy-related myeloid neoplasms (t-MNs). However, this medical history-oriented definition has an inherent uncertainty considering possibilities of pathogenetically only minimally related or even irrelevant to previous exposure to CT/RT in a part of patients. We investigated whether there were any specific factors that could separate 'by definition tMNs' into better vs. poorer prognosis, thereby necessitate different therapeutic strategies.

From July 2006 to May 2017, 82 adult patients were diagnosed as MDS or AML in Seoul National University Hospital, Seoul, Korea, with clear medical records of prior exposure to ≥ 1 of followings; alkylating agents, topoisomerase 2 inhibitors (Topo2I), and radiotherapy. After excluding 8 patients with acute promyelocytic leukemia, 74 patients (43 t-AML and 31 t-MDS) were analyzed (median age 58, range 19-82). The patients were treated by conventional therapeutic modalities for MDS/AML including intensive chemotherapy (IC; n = 30), hypomethylating agents (HMAs; n = 26), upfront allogeneic stem cell transplantation (allo-SCT; n = 2), low-dose cytarabine (n = 1), and best supportive care (BSC; n = 15). Lymphoid malignancies (31.1%) and breast cancer (21.7%) were the most common preceding cancers. More than half of the patients (54.1%; 40 patients) had received both alkylators and Topo2Is and previously known differences of characteristics between alkylator/radiation vs. Topo2I groups including median time of latency and initial presentation of t-MNs (either t-MDS or t-AML) were not observed. Median time from the last exposure of CT/RT to diagnosis of t-MN was 21.8 months (range, 0.8-134).

During median follow-up period of 20.6 months (95% CI 5.7-35.5 months) from diagnosis of tMNs, median overall survival (OS) was 11.6 months (95% CI 7.8-15.3 months). There was no difference of OS between t-MDS and t-AML (p = 0.136) or treatment with IC vs. HMAs (p = 0.143). Patients who had ≥ 3 abnormalities and/or complete or partial deletion of chromosome 5 or 7 (n = 35) showed inferior OS compared to those without the abnormalities (median OS 6.9 vs. 19.5 months; p = 0.048). By contrast, patients who had normal karyotype (NK) or core-binding factor (CBF) cytogenetics [t(8;21), inv(16), or t(16;16)] (n = 24; 16 for NK and 8 for CBF) showed significantly superior OS compared to their counterpart (median OS 38.8 vs. 6.2 months and 3-year OS 56.4% vs. 10.6%; p < 0.001; Figure). In addition to the absence of NK/CBF cytogenetics, age ≥ 70 years at diagnosis of t-MN, development of t-MN < 12 months from the last exposure to CT/RT, and metastatic or active status of preceding cancer at the time of t-MN diagnosis, were associated with inferior OS in univariate analysis. Multivariate analysis showed that absence of NK/CBF cytogenetics in addition to development of t-MN < 12 months from the last exposure to CT/RT, a parameter indicative of more recent damage of body by exposure to CT/RT, were independent factors of poor prognosis (Table).

Among 59 patients after excluding 15 patients who received BSC only for palliative aim, poor prognosis of lacking NK/CBF cytogenetics was overcome by allo-SCT (hazard ratio 0.44, 95% CI 0.20-0.95, p = 0.036). When analysis was limited to patients with t-AML who received either IC or HMAs, patients with NK/CBF cytogenetics had higher rate of achieving a complete remission (CR) compared to those without NK/CBF cytogenetics (75% vs. 25%). patients who achieved CR showed higher 3-year OS compared to those who had persistent disease (64.3% vs. 16.2%, p = 0.030).

Those results show that t-MN patients who have NK/CBF cytogenetics and fit to intensive treatment including allo-SCT can achieve comparable outcomes to de novo MDS/AML and suggest that those patients may have a different pathogenesis from that of genetically heavily damaged t-MNs. Future studies serially comparing genetic pathways and mutations within the heterogeneous tMNs would contribute to establish more fine-tuned, a genetics-oriented definition of tMNs.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.