Existing prognostic models for myelofibrosis (MF) are based on complete blood count parameters and clinical symptoms at the time of initial presentation, with the most widely cited being the Lille score (Blood 88.1013). These models, however, are only applicable to patients at the time of initial diagnosis, and they do not identify the patients at imminent risk of death. In order to identify patients in advanced phase (AP) MF that may be candidates for intensified therapy, we reviewed the records of 371 MF patients for disease features that predict for survival less than twelve months. Median baseline characteristics (range): age 63 yrs (24–86); male 59%; post-PV or post-ET MF 21%; spleen 5cm (0–30); HB <10g/dL or transfusion dependent (TD) 53%; abnormal cytogenetics 126/337 (37%). Patients presented at a median of 4 months (range, 0 to 296 months) from initial diagnosis; 6% were previously treated with alkylators or radiation, and 36% with hydroxyurea. Median follow-up was 46 months for survivors. Independent predictors of poor survival at initial presentation were: abnormal chromosome 17 (ABN17; p<0.001); age (p<0.001); hepatomegaly ≥5cm (p<0.001); platelets <100 (p<0.001); HB <10 or TD (p<0.001); abnormal WCC (p=0.007); PB or BM blasts ≥10% (p=0.008); splenomegaly ≥20cm (p=0.02); and male gender (p=0.02). Factors significant on univariate analysis but not significant on multivariate modelling included primary vs secondary MF, other cytogenetic abnormalities, previous radiation / alkylator or erythropoietin exposure, monocytes ≥1×10^9/L, abnormal creatinine or bilirubin, and performance status. Three disease features were chosen as criteria for AP because of their association with median survival <12 months: PB/BM blasts ≥10% (n=19, survival 10 months), ABN17 (n=9, survival 5 months) and severe cytopenia not related to therapy (PLT <25 and WCC <4.0, n=9, survival 8 months). These criteria were independent of existing prognostic models: among patients with a high-risk Lille score, median survival was significantly inferior for patients with AP criteria (n=19, 5 months) compared to those without (n=55, 24 months p<0.0001). In order to validate these criteria, we followed the progress of 334 patients not in AP at presentation. During follow-up, 56 additional patients met AP criteria, and median survival from AP development were: PB/BM blasts ≥10% (n=38), 12 months; ABN17 (n=6), 6 months; severe cytopenia not related to therapy (n=12), 10 months. The actuarial risk of AP development was 23% at 4 years. Of the 93 patients in AP, blast phase (BP) occurred in 30 (32%), and a further 46 (49%) died without reaching BP. In contrast, BP occurred in only 9 of 278 (3%) of patients in chronic phase (p<0.0001). Our observations indicate that PB/BM blasts ≥10%, ABN17 and severe cytopenia evolve during the natural history of primary or secondary MF and are associated with a high risk of blastic transformation or death from progressive disease. Recognition of such patients as AP is an important step in facilitating the development of high-risk intervention strategies.

Author notes

Disclosure: No relevant conflicts of interest to declare.