Prognostic risk models for transplant decision-making in myelofibrosis.

Prognostic models are widely used in clinical practice for transplant decision-making in myelofibrosis (MF). We have compared the performance of the International Prognostic Scoring System (IPSS), dynamic IPSS (DIPSS), and DIPSS-plus in a series of 544 patients with primary or secondary MF aged ≤ 70 years at the time of diagnosis. The median projected survival of the overall series was 9.46 years (95% confidence interval 7.44-10.59). Median survival for the highest risk groups was less than 4 years in the three prognostic models. By contrast, the projected survival for patients in the intermediate-2 categories by the IPSS, DIPSS, and DIPSS-plus was 6.6, 5.6, and 6.5 years, respectively. The number of patients in the intermediate-2 and high-risk categories was smaller in the DIPSS than in the IPSS or the DIPSS-plus. The IPSS and DIPSS-plus were the best models to discriminate between the intermediate-1 and intermediate-2 risk categories, which is a critical cut-off point for patient selection to transplant. Among patients assigned at diagnosis to the intermediate-2 or high-risk groups by the IPSS, DIPSS, and DIPSS-plus, only 17, 21, and 20%, respectively, were subsequently transplanted. In conclusion, in our contemporary series of younger MF patients only the highest risk categories of the current prognostication systems have a median survival below the 5-year threshold recommended for considering transplantation. Patient selection for transplantation can significantly differ depending on which prognostication model is used for disease risk stratification.