Longer Duration of Hydroxyurea Treatment in PV Increases Incidence of MF, Mortality

June 21, 2019
Darcy Lewis

A new meta-analysis quantifying the risks and benefits of using hydroxyurea to treat polycythemia vera found that while the incidence of thrombosis and acute myeloid leukemia were stable over time, mortality and transformation to myelofibrosis varied depending on follow-up duration.

A new meta-analysis quantifying the risks and benefits of using hydroxyurea (Hydrea) to treat polycythemia vera (PV) found that while the incidence of thrombosis and acute myeloid leukemia (AML) were stable over time, mortality and transformation to myelofibrosis (MF) varied depending on follow-up duration.

Published inHaematologica, authors led by Alberto Ferrari, MD, FROM Research Foundation, Papa Giovanni XXIII Hospital in Bergamo, Italy, wrote that thrombosis rates were 1.9% at a median patient age of 60. Those rates rose to 3.6% when the median age was 70 and 6.8% at a median age of 80. Previous cardiovascular complications predicted a higher incidence of arterial events.

Additionally, the leukemic transformation incidence rate was 0.4%. The authors found that both the incidence of transformation to MF and mortality significantly depended on patient age and follow-up duration. The MF rates were 5.0% at 5 years and 33.7% at 10 years. Similarly, the overall mortality rate was 12.6% at 5 years and 56.2% at 10 years.

“Hydroxyurea is the standard treatment in high-risk patients with polycythemia vera. Yet, estimates of its effect in terms of clinical outcomes (thrombosis, bleeding, hematological transformations and mortality) are lacking,” wrote Ferrari et al. “We provide reliable risk estimates for the main outcomes in [patients with] polycythemia vera under hydroxyurea treatment. These findings can help design comparative clinical trials with new cytoreductive drugs and prove the feasibility of using hard endpoints for efficacy, such as major thrombosis.”

The authors chose 16 articles (14 studies and 2 conference abstracts) to review whose design allowed them to retrieve the incidence of significant clinical outcomes in patients with PV. Most of the studies chosen were conducted in Europe, and some involved multiple countries.

The final meta-analysis examined outcomes for a total of 3236 patients. The smallest sample size was 25 patients, and the largest was 890. Length of follow-up ranged from 0.3 years to 12.4 years.

In looking at thrombosis, Ferrari et al found that the average incidence rate was 3.3%. This ranged from 1.9% at a median of 60 years in patients with no history of thrombosis to 6.8% at a median age of 80 years. They noted that estimates increased with the median age and were higher when there was a history of thrombosis, but the latter difference was not statistically significant. They performed a sub-analysis on arterial and venous thrombotic events and found that previous thrombosis was a highly significant predictor (P<.001) of arterial thrombosis incidence, but not of venous thrombosis.

Ferrari et al also examined hematological transformations and mortality. They found that the incidence of MF and overall mortality increased significantly after 5 years of follow-up. They estimated the risk of transformation to MF at a median age of 68 years to be 0.9% at 1 year, 5.0% at 5 years, and 33.7% at 10 years. The corresponding mortality risks under the same circumstances were 2.4% at 1 year, 12.6% at 5 years, and 56.2% at 10 years. The patient’s age at the start of follow-up determined whether their mortality risks would increase or decrease. The odds of MF transformation increased an average of 6% for each year of age (95% CI, 1%-11%). Mortality risks increased by 21% for each year of age (95% CI, 9%-33%).

The incidence of AML transformation was stable over time. The authors used a negative binomial model to calculate an annual rate of AML transformation of 0.4%. Their data suggested a slight increase in transformation risk after about 8 years.

Ferrari et al found the overall thrombosis rate to be about 3% per year, which they calculated by pooling each study’s event rate. They found that 2552 patients experienced 469 events, leading them to estimate a thrombosis incidence rate of 1.6%, in patients with a median age of 60. With a median age of 70, this rate rose to 3.6%, while it reached 6.8% with a median patient age of 80. “Contrary to what is widely known, we did not find a statistically significant effect of history of thrombosis on incidence of new vascular events,” they wrote. “…however, we highlight that the residual incidence of thrombosis in HU-treated [patients with] PV is still elevated, corresponding to approximately 10-fold higher than the one estimated in the general population.”

The authors found that the numbers of major bleeds and second cancers were too low to calculate meaningful incidence rates.

Overall, while Ferrari et al considered their estimates to be robust based on their large sample size, they noted the limitation that most of the studies they analyzed did not specifically address their study questions. “Some known predictors of clinical outcomes, such as history of thrombosis, which is a well-known risk factor for recurrences, turned out to be not-significant in meta-regression,” they wrote. “This may suggest that under HU treatment history of thrombosis is not a risk factor for recurrences anymore; but it may also be a byproduct of using aggregate data as predictors, with subsequent loss of information on individual patients.”

They believe their reference value can provide investigators with guidance in calculating necessary sample sizes for future comparative trials. “Lastly, we underline the value of an old, cheap, and safe molecule as a reliable and accessible resource for those settings where there is need to reconcile economic sustainability with the right to a quali-quantitative life advantage,” the study authors wrote.


Ferrari A, Carobbio A, Masciulli A, et al. Clinical outcomes under hydroxyurea treatment in polycythemia vera: a systematic review and meta-analysis [published online May 23, 2019].Haematologica.doi: 10.3324/haematol.2019.221234.