Differences in Clinical and Molecular Characteristics and Outcome in Prefibrotic and Overt Primary Myelofibrosis According to 2016 WHO Criteria. a Study on 639 Patients of the Agimm Group

The revised 2016 World Health Organization (WHO) classification of myeloid neoplasms dictated distinct criteria for prefibrotic (prePMF) and overt primary myelofibrosis (PMF), based on bone marrow (BM) morphology, including fibrosis grade (G)<1 in prePMF and G2-3 in PMF, and presence of leukoerythroblastosis in PMF.

AIM: to describe the characteristics and outcome of patients (pts) with a diagnosis of prePMF versus PMF according to the 2016 WHO criteria.

METHODS. We used a database of ca 800 pts collected in 5 Italian tertiary centers of the AGIMM project. Pts annotated with diagnosis of pre/early PMF and PMF were identified, and BM biopsies were re-classified according to current criteria. A total of 639 pts with full information were retrieved; all were annotated for both driver mutations (JAK2V617F, MPLW515x, CALR) and High Molecular Risk mutations (HMR; Vannucchi et al, Leukemia 2013;1861), including ASXL1, EZH2, SRSF2, IDH1/2.

RESULTS. Of the 639 pts, 274 (42.8%) were re-classified as prePMF and 365 (57.2%) as PMF. After a median follow-up (FU) of 3.6y, 212 pts (33.2%) had died, 23% prePMF vs 40.8% PMF (P<.0001); 69 pts (10.8%) transformed to leukemia (AL), 8.0% vs 12.9% in prePMF vs PMF (P= .033). At diagnosis, compared to PMF, prePMF pts were enriched in females (57.2% vs 42.9%, P=.01), were younger (59.4 vs 64.7y; P<.001) and with less >65y old individuals (36.1% vs 46.8%, P= .004), had higher Hb (12.7 vs 10.7g/dL; P<.0001) and fewer anemic subjects (Hb<10g/dL:14.2% vs 38.4%; P<.0001), showed higher platelet (plt) count (453 vs 247x10⁹/L; P<.0001) and fewer thrombocytopenic subjects (<100x10⁹/L; 8.0% vs 18.1%; P<.0001); blasts >1% were found in 12.0% vs 26.3% (P<.0001), while leukocyte count and % of pts with >25x10⁹/L were similar. Abnormal karyotype in 18.5% prePMF vs 38.6% PMF (n=317; P<.0001). Constitutional symptoms were reported in 20.8% prePMF vs 34.3% PMF (P<.0001), palpable splenomegaly in 64.2% vs 83.1% (P<.0001), spleen >10cm in 10.5% vs 24.1% (P<.0001). Major thrombosis occurred in 15.5% of prePMF vs 9.0% PMF (P=.02). According to IPSS, 74.2% and 25.8% of prePMF pts were in the lower and higher risk categories, respectively, vs 50.8% and 49.2% of PMF (P<.0001). At the latest FU, most prePMF pts maintained a lower risk category (65.8%) according to DIPSS unlike PMF with 69.3% being categorized as higher risk (P<.0001).

The proportion of patients with JAK2V617F (and their median allele burden), MPLW515x and CALR (type I and type II) mutations was similar in prePMF and PMF; triple-negative (TN) pts were slightly more frequent in PMF (14%) than prePMF (9.9%; P=.041). Conversely, 24.8% of prePMF pts vs 41.1% PMF were HMR (P<.0001), the frequent mutated genes were ASXL1 and EZH2; >2 HMR mutations, that are prognostically negative (Guglielmelli et al, Leukemia 2014; 28), were found in 11.8% of PMF pts vs 4.7% prePMF (P<.0001).

Median survival (OS) was 17.6y in prePMF vs 7.2y in PMF (P<.0001). OS was accurately predicted by IPSS in both prePMF and PMF. Using CALR type 1 mutation as the reference group, CALR type 2, JAK2/MPL mutations and triple negativity were negative predictors. HMR status was prognostically significant for OS in both prePMF (HR1.8, 95%CI 1.1-3.0, P=.03) and PMF (HR 2.5; 1.8-3.4, P<.0001) as it was >2 HMR mutations (HR 9.3, 4.5-19.0 and 3.4,2.1-5.3, respectively; P<.0001).

CONCLUSIONS. This analysis of pts with contemporary diagnosis of prePMF and PMF disclosed important clinical, hematologic and molecular differences between the two, and indirectly suggested that they might represent a phenotypic continuum where increased grade of fibrosis associates with worsening of disease manifestations and outcome.