Long-Term Outcome in Polycythemia Vera (PV): Final Analysis of a Randomized Trial Comparing Hydroxyurea (HU) to Pipobroman (Pi).

Background: Only 4 randomized clinical trials in PV have directly compared cytoreductive therapies (32P, busulfan, chlorambucil, HU, and Pi), all published after a median follow-up of 10 years or less. Thrombosis was the main cause of mortality and morbidity in all trials. Cumulative incidence of evolution to myelofibrosis (MF), and acute myeloid leukemia or myelodysplastic syndromes (AML/MDS) was about 8%, and 6%, respectively. In 1 of those trials, conducted between 1980 and 1996 by the French Polycythemia Study Group (FPSG) in 292 PV patients (pts) younger than 65 yrs, no difference between HU and Pi was found in terms of survival, risk of thrombosis, and AML/MDS/MF evolution at the time of first analysis with a median follow up of 7 years (

Methods: We updated the results of the FPSG trial at the reference date of April 15, 2008, with a median follow-up of 16.3 years. After randomization in the trial (136 pts in the HU arm, 149 in the Pi arm), 94 (33%) pts had received only HU, 130 (46%) only Pi, and 61 (21%) both drugs (including 42 who had switched from HU to Pi, and 19 from Pi to HU) during the whole follow-up period (7 patients were excluded from the final analysis because of incomplete follow up data). To take into account treatment crossovers, statistical analysis was performed both in “intention to treat” (ITT) and according to the main treatment received (i.e. treatment effectively received).

Results: Median survival was 20.3 years (95%CI: 16.4 – 25) in HU arm, and 15.4 years (95%CI: 13.4 – 17) in Pi arm (p=0.008). 95 pts have died, the 3 main causes of death being evolution to AML/MDS in 51 pts (54%), vascular events in 19 pts (20%), and solid tumor in 11 (12%) pts. The 51 cases of evolution to AML/MDS included 10 MDS, and 41 AML (including 5 preceded by an MDS phase). Cumulative incidence (CI) of AML/MDS at 10, 15, and 20 years in ITT analysis was 6.6%, 16.5%, and 24% in the HU arm, and 13%, 34%, and 52% in the Pi arm, respectively (p= 0.004). As the median duration of HU treatment (12 years) was significantly longer than that of Pi treatment (9.5 years, p=0.0002), we also performed the analyses according to the main treatment actually received by pts, which showed cumulative incidence of AML/MDS of 7%, 14%, and 22% with HU, and 12%, 37%, and 56% with Pi at 10, 15, and 20 years, respectively (p=0.008), confirming the results obtained in ITT analyses. Regarding MF, the CI at 10, 15 and 20 years was comparable in both arms in ITT analysis: 12.6%, 19%, and 27% in the HU arm, and 7.8%, 16%, and 27% in the Pi arm, respectively (p=0.7). However, a significantly higher incidence of MF was found in pts who had received HU as main treatment: 15%, 24%, and 32% at 10, 15, and 20 years, compared to 5%, 10%, and 21%, respectively, in patients who received mainly Pi (p=0.02).

Conclusion: Although medium-term analysis of FPSG trial did not show differences between HU and Pi, current update after a median follow-up of 16.3 years finally showed that median survival was shorter, and incidence of AML/MDS higher in Pi treated pts. By contrast, risk of MF was higher in pts treated predominantly with HU. Of note, was the clearly higher than previously reported cumulative incidence of AML/MDS with HU. Evolution to AML/MDS represented by far the first cause of death in the long-term (54% versus 20% for vascular events). Those results suggest that Pi should not be used as first line therapy in PV patients, and that very long term may be required to draw conclusions regarding leukemic evolution in PV trials.