Therapy-related myelofibrosis does not appear to exist

Therapy-related myeloid neoplasms, such as myelodysplastic syndromes (t-MDSs) and acute myeloid leukemia (t-AML), account for 10% to 20% of all myeloid neoplasms and are recognized as a distinct subgroup according to 2008 World Health Organization (WHO) criteria.¹  Although t-MDS/t-AML are often linked to previous exposure to topoisomerase II inhibitors, alkylating agents, or radiation,²  this distinction has been eliminated in the revised WHO classification.³  Patients with t-MDS/t-AML usually have poor clinical outcomes and often have specific cytogenetic abnormalities involving chromosomes 5, 7, 11q23 (MLL gene), or 21q22 (RUNX1/AML1 genes).^4,5  Although an association between myelofibrosis (MF) and a prior malignancy has been rarely reported,⁶  the existence of therapy-related MF (t-MF) as a complication of cytotoxic therapy or radiation for a prior other malignancy is unknown, and the therapy-specific impact on the clinicopathological presentation or outcome of subsequent MF has not been studied. Here, we analyzed the clinical characteristics and outcomes of patients with MF who had a prior malignancy to investigate whether t-MF exists.

We retrospectively searched the medical records of 1100 patients referred to our institution with a diagnosis of MF between 1984 and 2013. MF diagnosis was confirmed by 2008 WHO criteria.⁷  Clinical, demographic, and cytogenetic data were abstracted from the medical records. Direct sequencing of exons 12 and 13 of the ASXL1 gene was performed as described previously.⁸  Unfavorable cytogenetics at the time of MF diagnosis included chromosome 5, 7, and 17 abnormalities, trisomy 8, 11q23 abnormalities, or complex karyotype (>3 abnormalities).^4,9,10 

Chemotherapy (CHT) agents were classified by their mechanism of action and divided into the following categories: alkylating agents (eg, cyclophosphamide, bendamustine, chlorambucil, platinum derivatives), topoisomerase 2 inhibitors (eg, podophyllotoxin derivatives, anthracyclines); antimetabolites (eg, 5-fluorouracil, methotrexate, gemcitabine, hydroxyurea, cytarabine, cladribine) and antitubulin agents (eg, vinca alkaloids, taxanes).

Leukemic transformation was defined as blasts ≥20% in the peripheral blood or bone marrow. Overall survival (OS) was defined from the date of MF diagnosis to the date of death or last follow-up. The latency interval to MF was defined as the time from the date of prior other malignancy diagnosis to the date of MF diagnosis. Fisher exact, Mann-Whitney, and Kruskal-Wallis tests were used to compare categorical and continuous variables, respectively. OS was evaluated by Kaplan-Meier analysis and comparisons were made with the log-rank test. This study was approved under a chart review protocol approved by the institutional review board at The University of Texas MD Anderson Cancer Center and was conducted in accordance with the Declaration of Helsinki.

Of 1100 patients who were referred to our institution with a diagnosis of MF, we identified 121 patients (11%) who had been diagnosed with a preceding other malignancy. Of 121 patients, 23 (19%) had >1 prior malignancy, accounting for a total of 151 prior malignancies. Median age at the time of MF diagnosis was 70 years (range, 35-89 years). Median latency time was 86 months (range, 3-563 months). Median follow-up after MF diagnosis was 31 months (range, 0.1-213 months). At the date of last follow-up, 61% of patients (74 of 121) had died. Median OS was 45 months (range, 0.2-156 months). Causes of death were recorded for 33 patients (27%) and included MF/AML (n = 11), other malignancy (n = 5), multiorgan failure or complications from stem cell transplantation (n = 7), infection (n = 3), and other causes (n = 7). Seventeen patients with unknown cause of death had stable or early MF at the time of last follow-up (<12 months before death). The remaining patients were lost to follow-up and the cause of death is not known.

The most common prior malignancies were prostate (30%), melanoma (17%), and colorectal carcinoma (10%) (Table 1). In 25 patients (12%), the prior malignancy was hematologic. Seventeen patients (14%) were treated with single radiation therapy (XRT), 16 (13%) with single cytotoxic CHT, and 9 (7%) with both (CMT). Among the 25 patients exposed to CHT, 19 (76%) received either an alkylating agent or topoisomerase II inhibitor. The majority of patients received concurrent therapy (CHT, XRT, or CMT) as part of the initial regimen; only 5 received sequential therapy for 2 different prior malignancies.

Outcomes of patients who received CHT, XRT, or CMT for their prior malignancy (defined as the t-MF group) were compared with those receiving only surgery (SX) or a conservative approach (observation or hormonal treatment). Patients treated with XRT were significantly older at the time of MF diagnosis than all others with a prior malignancy (median age, 74 vs 69 years; P < .05), but no other differences in demographics or clinical characteristics were noticed across treatment groups. The median latency time to t-MF was 7 years (84 months; range, 12-528 months) and was similar regardless of the type of prior malignancy (hematologic vs solid tumor), treatment approach (CHT vs XRT vs SX), or type of chemotherapeutic agent (alkylating agent vs topoisomerase II inhibitor vs their combination vs others). Similarly, median OS did not differ by treatment approach or type of therapy.

Thirty-nine of 121 patients (32%) with a prior malignancy had detectable cytogenetic abnormalities (Table 2): 23 of 39 (59%) had a favorable karyotype (del20q or del13q).¹¹  Five patients had either chromosome 5 or 7 abnormalities after exposure to alkylating agents or radiation. Neither of the 2 patients with a complex karyotype had received CHT. No unique patterns of cytogenetic abnormalities were observed between treatment groups, malignancy types, or CHT agents.

Because t- AML/t-MDS have been shown to have a higher frequency of molecular mutations, such as in the histone modifier ASXL1,¹²  we evaluated the frequency of ASXL1 mutations in patients previously exposed to CHT or XRT. Samples collected at MF diagnosis were available for 33 of 42 patients exposed to CHT or XRT. ASXL1 mutations were detected in 14 of 33 patients (42%), and frequencies were similar regardless of previous therapy exposure (3 of 5 with CMT; 6 of 15 with CHT and 5 of 13 with XRT; Table 2). ASXL1 mutations were not associated with any specific type of prior malignancy (3 had hematologic and 11 solid malignancies). Similarly, the latency time did not differ among those exposed to different types of therapy or those with and without an ASXL1 mutation (data not shown).

Fifteen of 121 patients (12%) (median age at MF diagnosis, 76 years; range, 54-87 years) had progression to AML. Median time to AML progression from MF diagnosis was 2.5 years (range, 0.5-9 years). Four patients received CHT, 1 XRT, and 10 conservative treatment. The proportion of patients with AML progression was similar in the t-MF and conservative treatment groups (11% and 21%, respectively), and only 2 patients with AML progression had a high-risk karyotype.

In summary, this is the first reported systematic analysis of the role of therapy for a prior malignancy in patients with MF. Although we have observed a few cases of chromosomal abnormalities typically reported in patients with t-AML or t-MDS, we did not find a correlation between them or the ASXL1 mutation and previous therapy. Our findings suggest that the treatment regimen used for the prior malignancy is not associated with the development, latency interval, or prognosis of MF. t-MF does not appear to be a separate clinical entity and should not be recognized as a type of therapy-related myeloid neoplasm. Therefore, treatment should be focused on controlling the more aggressive disease. For example, a patient with early MF with normal counts and minimal symptoms should be fully treated for the other malignancy regardless of the MF diagnosis. Because the treatment options for MF remain limited, stem cell transplantation should be strongly considered for all eligible patients with MF and a second malignancy in remission, or a coexistent hematologic malignancy.

Given the small number of patients, especially with previous CHT or XRT exposure, our findings do not allow us to make definitive conclusions and should be validated in a larger patient population.

This work was supported in part by the National Institutes of Health, National Cancer Institute through Anderson Cancer Center Support Grant CA016672.

Contribution: L.M., J.E.C., S.V., Z.E., and H.M.K. participated in the patient’s care; L.M., G.T., K.J.N., Z.E., and S.V. wrote the paper; G.T. and T.M. performed the DNA sequencing; and all authors revised and approved the final manuscript.

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

Correspondence: Srdan Verstovsek, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd, Unit 428, Houston, TX 77030; e-mail: sverstov@mdanderson.org.