Erythropoiesis Stimulating Agents and the Risk of Leukemic Transformation in Primary Myelofibrosis.

BACKGROUND: Information regarding risk factors for leukemic transformation (LT) in primary myelofibrosis (PMF) is limited, although both leukocytosis and abnormal cytogenetics have previously been implicated. The current retrospective study examines clinical variables at diagnosis as well as specific treatment modalities for their effect on LT in PMF.

METHODS: Diagnoses of PMF and LT were based on World Health Organization criteria. In addition, study inclusion required availability, for review, of bone marrow histology and clinical data obtained at or within 6 months of diagnosis but prior to treatment intervention. Statistical methods used were standard and included Cox regression analysis of leukemia-free survival (LFS).

RESULTS: A master database of PMF patients seen at the Mayo Clinic from 1976 through 2006 was queried to identify 311 patients who fulfilled the aforementioned stipulated criteria for study inclusion; median age was 57 years and 41% were females. At presentation, 30% of the patients displayed a hemoglobin level (Hgb) of < 100 g/L, 20% a leukocyte count of < 4 or > 30 x 10⁹/L, 18% a platelet count of < 100 x 10⁹/L, 12% an absolute monocyte count (AMC) of ≥ 1 x 10⁹/L, 28% hypercatabolic symptoms and 35% a peripheral blood (PB) blast percentage of ≥ 1; 27 patients (9%) had a PB blast percentage of ≥ 3. Dupriez prognostic scoring system distribution for high, intermediate, and low risk disease was 30 (10%), 94 (30%), and 187 (60%) patients. Among 182 and 139 evaluable patients, 74 (41%) and 80 (58%) displayed cytogenetic abnormalities and JAK2V617F, respectively. At a median follow-up of 27 months (range 0–282), 27 cases (9%) of LT were documented at a median time from diagnosis to LT of 26 months (range 0.8–266). Effect of clinical and laboratory variables at diagnosis on leukemic transformation By univariate analysis, LFS was negatively affected by anemia (p=0.04), leukocytosis (p=0.04), PB monocyte count ≥ 1 x 10⁹/L (p=0.02), platelet count of < 100 x 10⁹/L (p=0.004), hypercatabolic symptoms (p=0.03) and PB blast percentage of ≥ 3 (p<0.0001); LFS was not affected by PB blast percentage of > 1 but < 3 (p=0.49). By multivariable analysis, significance was sustained only for PB blast percentage of ≥ 3 (p = 0.0002) and platelet count of < 100 x 10⁹/L (p = 0.02) with hazard ratios (HR) of 5.8 and 2.8 and 95% confidence intervals (CI) of 2.3–14.6 and 1.2–6.6, respectively. The presence of JAK2V617F did not affect LFS (p=0.98). Effect of specific treatment on leukemic transformation By univariate analysis, LT was significantly associated with history of splenectomy (p=0.01) and treatment with erythropoiesis stimulating agents (ESA; p=0.004), danazol (p=0.007), and androgens (p=0.03) but not with hydroxyurea (p=0.17), interferon alpha (p=0.95), thalidomide (p=0.26) or other drugs. On multivariable analysis, significance was sustained for both ESA (p = 0.005; HR 3.1, CI 1.4–6.8) and danazol (p = 0.01; HR 3.4, CI 1.3–8.5), even when the aforementioned prognostic indicators at diagnosis were added as covariates to the multivariable model.

CONCLUSIONS: The current study identifies PB blast percentage of ≥ 3% and platelet count of < 100 x 10⁹/L, at presentation, as independent risk factors for LT in PMF. Unexpectedly, the study also revealed an association between LT and treatment history with ESA or danazol. These observations are intriguing, considering recent reports on the possible harmful effects of ESA in certain solid tumors.