Background: The International Prognostic Scoring System (IPSS) (Blood 2009;113:2895), the dynamic(D) IPSS (Blood 2010;115:1703) and the DIPSS-plus (JCO 2011;29:392) are commonly used to predict survival among patients (pts) with Primary Myelofibrosis (PMF). These scores were developed using populations of PMF pts that differ from the 2 categories outlined in the 2016 WHO criteria, i.e. prefibrotic and overt PMF (Blood 2016; 127:2391) ; indeed, IPSS performed suboptimally when applied to the two PMF variants (Blood 2017; 129:3227) . The prognostic relevance of fibrosis grade and mutation profile, including both driver (JAK2/CALR/MPL) and other myeloid malignancies-associated genes (Blood 2017; 129:3227), has been shown. Furthermore, high molecular risk category (HMR), including pts with any one mutated genes of ASXL1, SRSF2, EZH2, IDH1/2 (Leukemia 2013; 27:1861), and unfavorable karyotype (JCO 2011;29:392) , both provided IPSS/DIPSS-independent prognostic information for overall survival (OS) and leukemia-free survival (LFS). The aim of current study was to develop an updated prognostic score that included molecular (MIPSS70) and, when available, cytogenetic information (MIPSS70-plus), specifically directed to pts with pre-PMF and overt-PMF 70yr old and younger that are potential candidates to stem cell transplantation (SCT).
Methods: Previously published methods were used to sequence JAK2, MPL, CALR, EZH2, ASXL1, IDH1/2 and SRSF2 . Survival was calculated from date of diagnosis (AGIMM cohort) or date of first referral (Mayo cohort). A Cox model with a stepwise selection procedure was used to select covariates significant for OS. The prognostic model was developed based on the magnitude of the hazard ratios (HR) in the training cohort.
Results: Development of MIPSS70. The learning cohort included 490 pts aged <70y (225 Pre-PMF, 265 overt-PMF) from 6 Italian institutions (AGIMM group). During a median FU of 5.8y (6.8y pre-PMF, 5.22y overt PMF), 25.3% of pre-PMF and 50.2% of overt-PMF pts died and 63 (9.8% pre-PMF, 15.5% overt PMF) progressed to leukemia. Mutation frequency was: JAK2 V617F 58%, CALR 26% (76% type1/like, 24% type2/like), MPL 5%, ASXL1 23%, SRSF2 7%, EZH2 6%, IDH1/2 2.7%; 30.8% were HMR-positive with 8% of pts presenting >2 HMR mutated genes. A HR-weighted score was assigned to variables maintaining significance in multivariable analysis: 2.0 points to leukocytes >25x109/L, platelets <100x109/L, >2 HMR mutations; 1.0 point to hemoglobin <100g/L, circulating blasts ≥2%, constitutional symptoms, fibrosis grade >2, absence of CALR type1/like, HMR category. Accordingly, 3 risk categories were delineated (Fig 1A): Low (score 0-1); Intermediate(score 2-4); and High (score >5) with 10-y survival of 83%, 39% and 12%, respectively. MIPSS70 yielded a better prediction of OS in a ROC analysis (MIPSS70 AUC = 0.760 vs. IPSS AUC = 0.710). Significantly different (p<0.0001) LFS was also predicted by the 3 risk categories. The aforementioned observations were validated in an external cohort of 223 patients from Mayo Clinic (details to follow at time of presentation). Finally, we showed that MIPSS70 efficiently resolved patients aged >70y in both cohorts.
Development of MIPSS70-plus.The learning cohort included 209 cytogenetically-annotated pts, aged ≤70y, from Mayo Clinic. Unfavorable vs favorable karyotype was added to the multivariable model that included all 8 MIPSS70 variables. Unfavorable karyotype (HR 2.6), presence of one (HR 1.7) or ≥2 (HR 2.3) HMR mutations, absence of type1/like CALR (HR 2.2), constitutional symptoms (HR 2.4), circulating blasts ≥2% (HR 1.7) and hemoglobin <10 g/dL (HR 1.5) remained significant. Based on HR-weighted scoring of these 6 variables, 4 risk categories were established and effectively predicted OS (Figure 1B) and LFS. These observations were validated in a separate cohort of 260 pts from the AGIMM group (details to follow at time of presentation). The MIPP70-plus was also effective in pts aged >70y.
Conclusions. The new MIPSS70 and MIPSS70-plus scores include modern disease-associated, risk variables pertinent to both pre-PMF and overt-PMF according to the 2016 WHO classification, and are intended to facilitate decision-making for patients 70yr old and younger who are potential candidates to SCT.
Vannucchi: Novartis: Honoraria, Speakers Bureau; Shire: Speakers Bureau. Rambaldi: Novartis, Amgen, Celgene, Sanofi: Other: Travel, Accomodations, Expenses; Novartis, Roche/Genentech, Amgen, Italfarmaco: Consultancy. Passamonti: Novartis Pharmaceuticals Corporation: Consultancy, Honoraria, Speakers Bureau.
Asterisk with author names denotes non-ASH members.