The gain-of-function mutation at codon 617 of JAK2 (JAK2V617F) is the most common somatic event observed in patients with polycythemia vera (PV), occurring in over 95% of PV patients. JAK2V617F confers cytokine hypersensitivity and cytokine-independent growth of erythroid progenitors, which are characteristic features of PV. Homozygous JAK2V617F is observed in about half of PV patients, whereas it is rarely seen in essential thrombocythemia (2-4%) and other myeloproliferative neoplasms. Homozygous JAK2V617F has been assumed to result from homozygous recombination, leading to uniparental disomy on 9p (9pUPD). It has been reported that the JAK2 46/1 (GGCC) haplotype may predispose carriers to the JAK2V617F mutation, and the JAK2V617F mutation facilities the acquisition of homozygous JAK2V617F. Challenging this view is a single study reporting 9pUPD in two PV subjects with wild-type JAK2, suggesting that in these two individuals, 9pUPD might have preceded the JAK2V617F mutation (Blood. 2011;118(24):6468-6470). However, the relationship between JAK2V617F and 9pUPD, the frequency of this new PV molecular subtype, its clinical relevance, and the stability of this genotype need to be systematically defined in a larger sample cohort.

To address this, we combined whole-exome sequencing (WXS) of DNA from 31 consecutive PV patients with high-resolution SNP arrays, and further validated our findings in two additional cohorts comprising 59 PV consecutive patients collected from a single institution. In addition, we investigated the stability of each molecular subtype by using serial samples collected from 25 PV patients. We obtained an average of 125x coverage on JAK2 locus by WXS (Illumina Hiseq2000) and 2225x coverage by targeted deep sequencing using Ion PGM sequencer.

Analysis of these data shows that the relationship between the JAK2 locus and 9pUPD is more complex than originally assumed. We defined 4 subgroups: 41% of patients had JAK2V617F in a heterozygous state without detectable 9pUPD (Subgroup I); 43% of patients had JAK2V617F with an allelic fraction in direct proportion to the level of 9pUPD (Subgroup II; homozygous JAK2V617F); 10% of patients harbored 9pUPD at approximately twice the level of the JAK2V617F allelic burden (Subgroup III; UPD with heterozygous JAK2V617F); and a small subset (6%) of patients exhibited trisomy of 9p, generating 3 copies of the JAK2 allele by chromosome duplication (Subgroup IV). No difference in the frequency of the JAK2 46/1 (GGCC) haplotype was found among these 4 subgroups. We found that this subtype classification was stable over time in over 60% of patients, whereas it transformed among the 9pUPD-positive subtypes in the remaining patients, indicating the outgrowth of a new PV subclone. While 2 PV patients with 9pUPD and wild-type JAK2 were previously reported (Blood. 2011;118(24):6468-6470), we now show a relative high proportion of PV patients having the novel, previously not recognized JAK2 genotype; i.e. JAK2 9pUPD with heterozygous JAK2V617F mutation. Our study will provide novel perspectives on the molecular basis of the evolution of PV and a better understanding of the roles of JAK2V617F and 9pUPD in this disease.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.