Polycythemia vera (PV) and essential thrombocythemia (ET) are myeloproliferative neoplasms characterized by a low rate of mutation events during follow-up. It has been reported that the acquisition of non-driver mutations plays a key role in disease progression. Since there is concern about the leukemogenic potential of hydroxyurea (HU) there is a need to define whether exposure to HU would promote the emergence of such mutations.

The objective of the present study was to analyze risk factors involved in the development of non-driver mutations in patients with PV and ET with long molecular follow-up.


We studied 100 patients (PV n = 63, TE n = 37) with a median follow-up of 10 years (1-13). Targeted NGS was performed using a panel of 51 myeloid-related genes in the most recent available sample and then investigated at diagnosis and through follow-up using serial samples with a 3-year interval. The probability of acquiring a non-driver mutation was estimated by the Kaplan Meier method using the log rank test for comparisons. Multivariate analyses of factors predicting for mutation acquisition were done by Cox regression.


The median age at diagnosis was 59 years (range 25-84). Seventy-seven patients were treated with HU for a median time of 4.25 years (range 0.1-27), 9 patients received busulfan, 6 IFN, 6 ruxolitinib, 3 P32. Twenty-four patients did not receive any cytoreductive treatment. Disease-progression to acute myeloid leukemia (AML) and myelofibrosis (MF) was documented in 12 and 24 cases, respectively.

Non driver mutations were detected at diagnosis in 35 patients, eleven of them (31%) developed new mutations during follow-up. In 13 out of 65 (20%) patients without non-driver mutations at first sample, acquisition of new mutations was observed at different serial samples.

The 10-year probability of mutation acquisition was 27% in the whole cohort, 16.4% in patients who remained in the chronic phase, 34.3% in those evolved to MF, and 69% in those evolved to AML. The median interval time elapsed between the first sample and the first mutation acquisition was 9.27 years (range 0.8-26.8).

In the univariate analysis, the presence of non-driver mutations at diagnosis (p = 0.008), the number of mutations (p <0.0001) and mutations in SRSF2 (p <0.0001), IDH1 / 2 (p = 0.045) and RUNX1 genes (p <0.0001) were associated with a higher probability of mutation acquisition. Length of HU exposure was not correlated with the acquisition of mutations (p = 0.794).

In multivariate analysis, the presence of non-driver mutations at first sample (HR: 2.91, 95% CI 1.2-7, p = 0.018), number of mutations (HR: 2.53, 95% CI 1.45-4.4, p = 0.001), SRSF2 mutations (HR: 12.7, 95% CI 2.1-77.6, p = 0.006) and RUNX1 mutations (HR: 33.59; 95% CI 4.8-232.7 p <0.0001) were associated with a higher risk of new mutation after correction by age, type of diagnosis and total duration of treatment with HU.


The presence and number of non-driver mutations confer genetic instability in PV and ET by favoring the appearance of new genetic events.

In the present cohort of patients we could not prove the genotoxicity of HU as measured by the acquisition of non-driver mutations.


Supported by the grants from the Instituto de Salud Carlos III, Spanish Health Ministry, FISPI13/00557, FISPI1300393, RD012/0036/0004, SGR567. Alicia Senín received a grant from the SEHH (Sociedad Española de Hematología y Hemoterapia).


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.