Molecular Predictors of Disease Progression to Myelofibrosis (MF) in Patients (Pts) with Polycythemia Vera (PV) Enrolled in Reveal

Introduction: Progression of PV to MF represents a major cause of morbidity and mortality; however, predictors of progression remain unclear. The Prospective Observational Study of Patients With Polycythemia Vera in US Clinical Practices Trial (REVEAL; NCT02252159) followed pts for a median of ~4 years, with optional biospecimen collection every 12 months. This analysis investigated the clonal architecture and molecular mechanisms that predict transformation to MF in pts with PV.

Methods: Of 2510 pts enrolled in REVEAL, 1880 had an available biospecimen and a confirmed JAK2 mutation by ddPCR. Of these pts, 114 had MF transformation during the study period based on modified World Health Organization criteria. Whole exome sequencing was performed on enrollment biospecimens (pretransformation) from the 114 pts, as well as a control cohort of 340 pts without transformation, matched (1:3) by propensity score based on age, sex, duration of PV, body mass index, and history of thrombosis. Sequencing was performed using the Illumina (paired-end) platform at an average depth of 300× and processed using Genome Analysis Toolkit best practices. Variants were filtered (VarSeq) according to the following criteria: ≥60 reads coverage, ≥3 reads support, >1% variant allele frequency (VAF), and <1% (or missing) minor allele frequency in the Genome Aggregation Database. Variants were annotated via tiers based on predicted pathogenicity using multiple algorithms. Data were analyzed using fast gene set enrichment analysis (FGSEA) and cell type-specific enrichment analysis (CSEA).

Results: There were no significant differences in clinical characteristics at enrollment between the transformed and nontransformed groups. Median (range) enrollment duration was 18.2 (0-30.0) and 19.1 months (0-32.5) in the transformed and nontransformed groups. A wide JAK2 V617F VAF distribution was observed in both cohorts; however, median (range) VAF was significantly higher in the transformed vs nontransformed group (84.21 [0.02-97.99] vs 57.45 [0.05-98.65]; P<0.0001). For both transformed and nontransformed pts, prevalence of commonly mutated genes was consistent with observations in other myeloid malignancies. A higher number of combined Tier I/II mutations in myeloid-related genes was observed per pt in the transformed vs nontransformed group (mean [range], 5.6 [2-14] vs 4.8 [1-12]; P<0.05). In particular, SF3B1, IDH1/2, EZH2, and TP53 mutations were enriched in the transformed group, with VAFs that trended higher compared with the nontransformed group. In addition, the presence of multiple clones was suggested by a difference in VAF between JAK2 V617F and co-occurring somatic variants.

To identify novel genes potentially contributing to PV transformation, genes with Tier I variants significantly associated with transformation (P<0.05) and an odds ratio >2 were prioritized. 83 genes were identified, including the tumor suppressor gene PTPN2, which has a known role in regulation of inflammation. Copy number variant (CNV) analysis identified 25 genes with gains and 28 with losses in the transformed vs nontransformed group. More pts in the transformed group had a CNV gain in TYRO3, which has been linked to myeloid leukemia cell growth. Pathways with the highest enrichment scores via FGSEA comparison of the transformed vs nontransformed group included Wnt/β-catenin signaling, inflammatory response, and DNA repair, whereas CSEA indicated a strong signal in bone marrow granulocytes and blood monocytes. Analysis of Tier II variants further identified markers of upregulated interferon-α response, notch signaling, and IL-6 and JAK-STAT3 signaling. Together, these data suggest a potential signature of inflammation in PV transformation to MF.

Conclusions: Although clinical characteristics were similar at enrollment for all pts in this analysis, JAK2 V617F VAF was significantly higher in pts with PV who transformed to MF. In addition, the VAF of nondriver somatic mutations trended higher in multiple genes with a known role in myeloid malignancy. Ongoing analyses of novel genes suggest a role of inflammatory pathways in PV transformation and may uncover additional pathways that contribute to this process. Further investigation will include analysis of clonal evolution via intra-pt longitudinal biospecimens and development of a machine learning approach to predict pts at increased risk of transformation.