Clinical characteristics and outcomes of therapy-related chronic myelomonocytic leukemia

Therapy-related myeloid malignancies are a distinct group of myeloid neoplasms that are associated with exposure to certain cytotoxic chemotherapies or ionizing radiation. Currently, the World Health Organization (WHO) defines 2 types of therapy-related myeloid neoplasms: (1) alkylating agent–related or radiation-related acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS); and (2) topoisomerase II inhibitor–related AML.¹  In general, these therapy-related diseases are associated with high-risk cytogenetic abnormalities and respond poorly to conventional therapies.

Chronic myelomonocytic leukemia (CMML) is characterized by persistent peripheral blood monocytosis (>1.0 × 10⁹/µL), a variable bone marrow blast of <20%, and dysplastic hematopoiesis.²  On rare occasions, CMML is diagnosed in patients who have been exposed to cytotoxic chemotherapies or ionizing radiation, hence considered as therapy-related CMML (t-CMML). Although several case reports of t-CMML have been identified,^3-7  no systematic analysis of the disease’s clinical characteristics and prognosis have been conducted owing to its extreme rarity.

The present study aims to describe the clinical features and outcomes of t-CMML patients and compare these characteristics with those of de novo CMML patients.

We identified 358 CMML patients who were referred to MD Anderson Cancer Center (MDACC) between January 2003 and July 2012. The median time from outside diagnosis to MDACC referral was 1.5 months (range: 0-80). Of those patients, 39 (11%) had prior exposure to chemotherapy and/or radiation therapy and were defined as t-CMML. Clinical data of the studied patients were obtained at the time of referral to MDACC. Therapies that were given to the patients were categorized into 4 groups: (1) best supportive care or cytoreductive therapy using hydroxyurea or oral etoposide (BSC/CR); (2) hypomethylating agents (HMA) such as 5-azacitidine–based or decitabine-based regimens; (3) AML-like induction therapy; and (4) other therapies including immunomodulatory drugs (thalidomide or lenalidomide) or tyrosine kinase inhibitors (iMiDs/TKIs). Cytogenetics and mutational analyses of the FLT3, KRAS, NRAS, and JAK2 genes were performed as described previously.^8-10  The study protocol followed the guidelines of the Declaration of Helsinki and was approved by the Institutional Review Board at MDACC.

Statistical methods and term definitions are described in the supplemental Methods; see the Blood Web site).

The demographics and clinical characteristics of 319 de novo CMML and 39 t-CMML patients are given in Table 1. Cytogenetics was categorized by CMML-specific cytogenetic risk classification.^11,12  The proportion of patients with intermediate- or high-risk cytogenetic abnormalities in the t-CMML group (50%) was significantly higher than that in the de novo CMML group (26%; P = .011). We used the MDAPS and CPSS to evaluate prognostic risk in the groups.^11,13  Among the evaluable patients, both scoring systems were verified to define distinct subgroups among the de novo CMML patients (supplemental Figures 1A, 2A, and 3A), and they appeared to identify high-risk patients in the t-CMML group despite the small cohort size (supplemental Figures 1B, 2B, and 3B). Although the distributions of MDAPS-defined risk groups did not differ significantly, the proportion of patients with CPSS-defined low risk in the t-CMML group (3%) was significantly smaller than that in the de novo CMML group (25%; P = .024) (Table 1). This discrepancy in risk distribution likely occurred because the CPSS incorporates cytogenetics in its score calculation, whereas the MDAPS is based on hematologic parameters only. De novo and t-CMML patients received similar types of therapies (P = .537), and the proportion of patients who underwent SCT was not statistically different between the 2 groups (P = .077) (Table 1).

The median latency to diagnosis of t-CMML was 6 years (range: 1-32) (Table 1). The most common primary cancers in the t-CMML group were lymphoid malignancies (37%), followed by prostate cancer (31%) and breast cancer (13%). None of the patients in t-CMML cohort had prior history of antecedent myeloid diseases. Fifteen (38%) patients were exposed to radiation therapy alone, whereas 24 (62%) were exposed to chemotherapy or combined modality therapy.¹⁴  Cytogenetic risk and CPSS risk showed a similar pattern between the radiation therapy alone and chemotherapy or combined modality therapy groups (supplemental Tables 1 and 2), and overall survival (OS) was similar between the 2 groups (P = .428) (supplemental Figure 3).

The median OS and leukemia-free survival (LFS) of the t-CMML patients (13 months [confidence interval, CI: 2.8-23.2] and 9 months [CI: 14.5-13.5], respectively) were significantly shorter than those of the de novo CMML patients (22 months [95% CI: 17.7-26.3] and 16 months [95% CI: 12.2-19.8], respectively; P = .023 and 0.011, respectively) (Figure 1). Survival differences by the therapy groups are shown in supplemental Figure 4.

Supplemental Table 3 shows the result of the univariate analyses for OS and LFS in the study group. The following variables were fitted into a multivariate Cox proportional hazard model: red blood cell transfusion dependency, WHO classification, CMML-specific cytogenetic risk, LDH, WBC, SCT, categorized therapies, and t-CMML. After adjustment for the prognostic effects of these covariates, t-CMML had no significant impact on OS or LFS (supplemental Table 4). The significantly shorter OS and LFS durations of the t-CMML patients in the univariate analyses were likely driven by high-risk cytogenetic abnormalities. In fact, when we removed the cytogenetic risk variable from the multivariate model, t-CMML showed statistically significant prognostic impact on OS and LFS (data not shown).

To the best of our knowledge, this is the first systematic study to show the clinical characteristics and outcomes of a reasonably large number of t-CMML patients treated at single institution. The results of the present study showed that although the groups have similar demographics and hematologic and molecular abnormality profiles, t-CMML patients possess more high-risk cytogenetic abnormalities than de novo CMML patients do. This is somewhat similar to what has been observed in patients with therapy-related AML or MDS, in whom a high incidence of poor-risk cytogenetic abnormalities is the distinguishing feature from de novo disease. In fact, OS of t-CMML was similar to that of therapy-related MDS patients who were treated at our institution during the same period (supplemental Figure 5).

In their literature review, Ahmed et al identified 8 t-CMML patients.³  Of the 5 patients who underwent cytogenetic testing, 4 carried the high-risk karyotype, and the median latency to t-CMML was 7 years (range: 2-14). These results are essentially consistent with the present study’s findings. The incidence of t-CMML in our analysis of more than 350 CMML patients was 11%, a number that perhaps reflects significant institutional bias as our institution is a tertiary cancer center.

Our multivariate model showed that HMA treatment was associated with better OS and LFS in CMML patients (supplemental Table 4). Subgroup analysis in the t-CMML cohort suggested that survival benefit by HMA treatment may be more evident in t-CMML patients (supplemental Figure 4). These findings need further validation in a larger cohort.

In summary, our findings indicated that patients can develop t-CMML 6 to 7 years after exposure to cytotoxic chemotherapy or ionizing radiation. t-CMML is associated with higher-risk cytogenetic abnormalities that manifest as poor outcomes. A large-scale multicenter study is warranted to confirm our findings. We propose that this rare disease be recognized as a new entity of therapy-related myeloid neoplasms.

The authors would like to thank Joseph Munch for his critical editing of the manuscript and Prof Akifumi Takaori at Kyoto University for providing significant intellectual input to the project.

This work was supported by the Ruth and Arnold Fund (G.G.-M.), the Edward P. Evans foundation (G.G.-M.), the Celgene Future Leaders in Hematology Award (K.T.), the Kimberly Patterson Leukemia Fellowship (K.T.), and in part by the National Institutes of Health through Anderson’s Cancer Center Support Grant CA016672.

Contribution: K.T. designed the study, collected data, analyzed data, and wrote the manuscript; N.P. treated patients and wrote the manuscript; P.S. collected data and approved the manuscript; G.N.-G. and J.N. analyzed the data and wrote the manuscript; R.L. conducted molecular analyses and approved the manuscript; C.B.-R. conducted pathological diagnosis and approved the manuscript; S.P. collected data and approved the manuscript; J.C. and H.K. treated patients, provided significant intellectual input, and approved the manuscript; and G.G.-M. designed the study, guided the project, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Guillermo Garcia-Manero, Department of Leukemia, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd, Box 428, Houston, TX 77030; e-mail: ggarciam@mdanderson.org.