- Anna L. Godfrey1,2,3,
- Edwin Chen1,2,3,
- Francesca Pagano1,
- Yvonne Silber1,3,
- Peter J. Campbell4 and
- Anthony R. Green1,2,3⇑
- 1Cambridge Institute for Medical Research and Department of Haematology, University of Cambridge, Cambridge
- 2Department of Haematology, Addenbrooke’s Hospital, Cambridge
- 3Wellcome Trust – Medical Research Council Cambridge Stem Cell Institute, University of Cambridge, Cambridge
- 4Wellcome Trust Sanger Institute, Hinxton, Cambridge, UK
Subclones homozygous for JAK2V617F are more common and larger in patients with polycythemia vera compared to essential thrombocythemia, but their role in determining phenotype remains unclear. We genotyped 4564 erythroid colonies from 59 patients with polycythemia vera or essential thrombocythemia to investigate whether the proportion of JAK2V617F -homozygous precursors, compared to heterozygous precursors, is associated with clinical or demographic features. In polycythemia vera, a higher proportion of homozygous-mutant precursors was associated with more extreme blood counts at diagnosis, consistent with a causal role for homozygosity in polycythemia vera pathogenesis. Larger numbers of homozygous-mutant colonies were associated with older age, and with male gender in polycythemia vera but female gender in essential thrombocythemia. These results suggest that age promotes development or expansion of homozygous-mutant clones and that gender modulates the phenotypic consequences of JAK2V617F homozygosity, thus providing a potential explanation for the long-standing observations of a preponderance of men with polycythemia vera but of women with essential thrombocythemia.
The JAK2V617F mutation is found in over 95% of patients with polycythemia vera (PV) and approximately 60% of those with essential thrombocythemia (ET).1 but the additional mechanisms which determine their distinct clinical phenotypes remain unclear. Circumstantial evidence suggests that JAK2V617F homozygosity may have a role in determining the PV phenotype: i) homozygous-mutant hematopoietic precursors are isolated from a higher proportion of patients and form larger clones in PV compared to ET;2–4 ii) mouse models have demonstrated PV-like, rather than ET-like, phenotypes associated with higher JAK2V617F gene dosage;5,6 iii) higher JAK2V617F allele burdens in granulocyte DNA have been associated with higher hemoglobin levels and lower platelet counts in PV.7,8
However, several lines of evidence indicate that the role of JAK2V617F homozygosity in PV pathogenesis is not simple. Clonal analyses indicate that, in fact, small homozygous-mutant clones arise frequently and recurrently in both PV and ET, suggesting that the mere presence of homozygosity is insufficient for a PV phenotype, and additional factors are required.4 Studies of some,9–11 but not other12 knock-in mouse models have reported that heterozygosity for JAK2V617F is associated with marked erythrocytosis, with no further increase in hemoglobin levels in homozygous mice.10 Associations of JAK2V617F allele burden with blood counts show inconsistencies between studies13 and have not determined the extent to which allele burdens reflect the relative proportions of heterozygous and homozygous-mutant cells. Moreover, reduced levels of STAT1 activation have been identified in heterozygous-mutant erythroblasts from patients with PV, compared to those with ET, and may contribute causally to the PV phenotype.14 It is, therefore, unclear whether JAK2V617F homozygosity makes a significant contribution to the pathogenesis of human PV.
Here we used a clonal approach to test the hypothesis that JAK2V617F homozygosity has a causal role in the PV phenotype, and investigated its associations with demographic features, by utilizing genotype data from 4564 erythroid colonies from 59 patients with PV or ET.
Design and Methods
Patients were recruited from Addenbrooke’s Hospital, Cambridge, UK. Demographic and clinical features, together with methods used for colony culture and genotyping, have been previously described.4 Patients met diagnostic criteria for JAK2V617F-positive PV or ET according to the British Committee for Standards in Haematology.15,16 The study was approved by the Cambridge and Eastern Region Ethics Committee, patients gave written informed consent, and research was carried out in accordance with the Declaration of Helsinki.
To test the associations of the relative proportions of JAK2V617F-homozygous and heterozygous colonies with hematologic or demographic features, Poisson’s regression analyses were performed for the count of JAK2V617F-homozygous colonies, with total JAK2V617F-mutant colonies as an offset. P<0.05 was considered significant.
Results and Discussion
In order to assess the importance of JAK2V617F homozygosity in hematologic phenotype, we first investigated the relationship between relative proportions of JAK2V617F-homozygous and heterozygous hematopoietic precursors and hematologic features in PV and ET. We used BFU-E colonies because overactive erythropoiesis is a feature of both PV and JAK2V617F-positive ET1 and is important in distinguishing between the two. Moreover, the proportions of colony genotypes are similar between BFU-E and CFU-G in PV patients.3 Colony genotypes are summarized in Online Supplementary Table S1. Homozygous-mutant erythroid precursors were identified in 24 of 30 patients with PV and 15 of 29 patients with ET. The ratios of homozygous-mutant to heterozygous colonies were generally higher in PV than ET, but varied significantly within each subgroup, and were reproducible for individual patients.4
A causal relationship between JAK2V617F homozygosity and PV would predict an association between a higher proportion of JAK2V617F-homozygous colonies and more extreme PV-like clinical features, even within a cohort of PV patients. We used Poisson’s regression analyses to ask whether the relative proportion of homozygous-mutant precursors, as a proportion of all mutant precursors, was associated with hematologic features at diagnosis. This was a retrospective analysis since blood counts at diagnosis were being compared with the proportions of homozygous- and heterozygous-mutant precursors at the time of colony assays. In univariable analyses of the combined PV and ET cohort, a larger proportion of homozygous precursors were associated with higher hemoglobin, higher white cell count and lower platelet count at diagnosis (all P<0.0001). In a multivariable analysis, controling for diagnosis, age, gender, disease duration, presence of cytoreductive therapy and erythropoietin concentration in colony assays, an increased proportion of homozygous colonies remained significantly associated with higher hemoglobin, higher white cell count and lower platelet count at diagnosis in the whole patient group and the PV subgroup, but not in the ET subgroup (Table 1). A larger proportion of homozygous colonies were, therefore, independently associated with more extreme PV-like features at diagnosis in the whole JAK2V617F-positive cohort and in the PV subgroup.
We considered the possibility that expansion of homozygous clones is secondary to the abnormal blood counts in PV, for example, through erythrocytosis suppressing erythropoietin levels and selecting for JAK2V617F-homozygous progenitors. However, multivariable analyses showed that a higher proportion of homozygous colonies was not associated with higher hemoglobin, higher white cell count or lower platelet count at the time of colony assays (data not shown). A higher proportion of homozygous-mutant colonies were, therefore, associated with more extreme PV-like blood counts at diagnosis but not at the time of colony assay, consistent with the concept that JAK2V617F homozygosity contributes causally to a PV phenotype.
In order to assess whether other clinical features predispose to or interact with JAK2V617F-homozygous clones, we next used Poisson’s regression analysis to test whether the number of homozygous-mutant precursors, as a proportion of all mutant precursors, was associated with demographic features or disease duration. In univariable analyses of the whole cohort, a larger proportion of homozygous colonies were associated with a diagnosis of PV (rather than ET), older age at assay, male sex and shorter time from diagnosis to assay (all P<0.0001). In a multivariable analysis, a larger proportion of homozygous colonies were associated with a diagnosis of PV and older age at time of assay (both P<0.0001, Table 2). A surprising interaction was identified between gender and diagnosis (P=0.003): an increased proportion of homozygous colonies were associated with male gender in PV (P=0.0006) but with female gender in ET (P=0.033). A significant interaction was also found between disease duration and diagnosis (P<0.0001), with an increased proportion of homozygous colonies associated with shorter disease duration in PV but with longer disease duration in ET (both P<0.0001). These associations were confirmed to be significant in multivariable analyses within the PV and ET subgroups (Table 2).
The association between an increased proportion of homozygous progenitors and increasing age, identified in both PV and ET, was also found in a cohort of 18 patients with JAK2 exon 12-mutated PV4 (P=0.024 in multivariable analysis) and was independent of disease duration. These data raise the possibility that age-dependent changes in DNA damage17 and/or DNA repair mechanisms18,19 contribute to an increased rate of mitotic recombination, and thus acquisition of JAK2 mutation homozygosity, in older patients. Alternatively, expansion of JAK2V617F-homozygous subclones may require additional mutations4 which accumulate with age.
The associations of an increased proportion of homozygous-mutant colonies with shorter disease duration in PV, but with increasing disease duration in ET, are likely to reflect distinct mechanisms in the two diseases. In ET, homozygous-mutant clones are present in approximately 50% of patients and are small.4 Our results indicate that, over time, mitotic recombination is more likely to occur and/or that an increasing proportion of ET patients develop clones above the detection threshold. By contrast, most PV patients have large homozygous-mutant clones.4 The association of an increased proportion of homozygous-mutant colonies with shorter time from diagnosis suggests that the size of homozygous-mutant clones, relative to heterozygous clones, decreases with time. This observation does not merely reflect suppressed erythropoietin levels and selection for homozygous-mutant precursors soon after diagnosis, since the association remained significant (P<0.0001) when we removed 3 patients with hemoglobin levels or hematocrit above the normal range at the time of colony assay, or when hemoglobin at time of assay was incorporated as a variable. Instead, we favor the possibility that hydroxycarbamide suppresses JAK2V617F-homozygous progenitors more than their heterozygous counterparts. This effect would be more evident in PV given the large number of homozygous-mutant colonies in most PV patients, and is consistent with evidence that hydroxycarbamide reduces granulocyte JAK2V617F allele burden especially in patients with PV and higher initial allele burdens.20,21
An association between male gender and increasing granulocyte JAK2V617F allele burden has been reported in PV, and was suggested to reflect gender-related differences in the frequency of mitotic recombination.22 Our data, however, indicate that the proportion of homozygous-mutant colonies was higher in males with PV than in females with ET, and do not favor this interpretation (Figure 1A). An alternative model is that JAK2V617F homozygosity occurs in both sexes, but that the phenotypic consequences of homozygosity are modulated by gender, so that the occurrence of a dominant JAK2V617F-homozygous clone is more likely to lead to development of PV in a male than in a female (Figure 1B). This would predict an excess of males with homozygosity in PV but an excess of females with homozygosity in ET, consistent with our data. The effect of JAK2V617F homozygosity on erythropoiesis could be modulated by gender-specific factors including a permissive effect of androgens in men and a constraining effect of iron depletion in pre-menopausal women. Importantly, this model provides an explanation for the long-standing observations that PV is more common in men23 whereas ET is more common in women.24
The authors would like to thank the Cambridge Blood and Stem Cell Bank for sample collection.
The Online Version of this article has a Supplementary Appendix.
Work in ARG’s laboratory is supported by Leukemia and Lymphoma Research, the Wellcome Trust, the Medical Research Council, the Kay Kendall Leukaemia Fund, the Cambridge NIHR Biomedical Research Center, the Cambridge Experimental Cancer Medicine Centre and Leukemia and Lymphoma Society of America. ALG is a Kay Kendall junior clinical fellow. PJC is a Wellcome Trust senior clinical fellow.
Authorship and Disclosures
Information on authorship, contributions, and financial and other disclosures was provided by the authors and is available with the online version of this article at www.haematologica.org.
- Received October 11, 2012.
- Accepted December 18, 2012.
- Copyright© Ferrata Storti Foundation