The impact of thrombosis on probabilities of death and disease progression in polycythemia vera: a multistate transition analysis of 1,545 patients

We applied a parametric Markov five-state model, on a well-characterized international cohort of 1,545 patients with polycythemia vera (PV; median age 61 years; females 51%), in order to examine the impact of incident thrombosis on the trajectory of death or disease progression. At a median follow-up of 6.9 years, 347 (23%) deaths, 50 (3%) blast phase (BP), and 138 (9%) fibrotic (post-PV MF) transformations were recorded. Incident thrombosis occurred at a rate of 2.62% pt/yr (arterial 1.59% and venous 1.05%). The probability of death, in the first 10 years, for 280 (18%) patients who developed thrombosis during follow-up was 40%, which was two-fold higher than that seen in the absence of thrombosis or any other transition state (20%; p < 0.01); the adverse impact from thrombosis was more apparent for arterial (HR 1.74; p < 0.01) vs venous thrombosis (p=NS) and was independent of other fixed (i.e., age, prior venous thrombosis, leukocytosis) or time-dependent (i.e., progression to BP or MF) risk variables. The transition probability to post-PV MF increased over time, in a linear fashion, with a rate of 5% capped at 5 and 10 years, in patients with or without incident thrombosis, respectively. The impact of thrombosis on transition probability to death or post-PV MF tapered off beyond 10 years and appeared to reverse direction of impact on MF evolution at the 12-year time point. These observations suggest thrombosis in PV to be a marker of aggressive disease biology or a disease-associated inflammatory state that is consequential to both thrombosis and disease progression.