Inherited predisposition to myeloproliferative neoplasms

Abstract

Myeloproliferative neoplasms (MPNs) are haematological disorders characterized by an overproduction of mature myeloid cells with a tendency to transform to acute myeloid leukaemia. Clonal proliferation of myeloid progenitor cells is driven by somatically acquired mutations, most notably JAK2 V617F, but there are important features relating to pathogenesis and phenotypic diversity that cannot be explained by acquired mutations alone. In this review we consider what is currently known about the role that inherited factors play in the development and biology of both sporadic and familial forms of MPN. Although most MPN cases appear to be sporadic, familial predisposition has been recognized for many years in a subset of cases and epidemiological studies have indicated the presence of common susceptibility alleles. Currently the JAK2 46/1 haplotype (also referred to as ‘GGCC’) is the strongest known predisposition factor for sporadic MPNs carrying a JAK2 V617F mutation, explaining a large proportion of the heritability of this disorder. Less is known about what genetic variants predispose to MPNs that lack JAK2 V617F, but there have been recent reports of interesting associations in biologically plausible candidates, and more loci are set to emerge with the application of systematic genome-wide association methodologies. Several highly penetrant predisposition variants that affect erythropoietin signalling, thrombopoietin signalling or oxygen sensing have been characterized in families with nonclonal hereditary erythrocytosis or thrombocytosis, but much less is known about familial predisposition to true clonal MPN. The heterogeneous pattern of inheritance and presumed genetic heterogeneity in these families makes analysis difficult, but whole exome or genome sequencing should provide novel insights into these elusive disorders.

Introduction

Myeloproliferative neoplasms (MPNs) are a group of related haematological disorders characterized by an overproduction of mature blood cells and a tendency to transform to acute myeloid leukaemia [Campbell and Green, 2006]. Although most cases appear to be sporadic, familial predisposition has been recognized for many years in a subset of cases and epidemiological studies have indicated the presence of common susceptibility alleles. This review discusses how inherited variants contribute to the development of both sporadic and familial forms of MPN.

General features of MPNs

Classification of classical MPNs is based on which myeloid cell lineage is predominantly expanded in the peripheral blood, namely elevated red cell mass in polycythaemia vera (PV), and elevated platelet numbers in essential thrombocythaemia (ET). Primary myelofibrosis (PMF) patients display bone marrow fibrosis, a variable myeloid cell number, extramedullary haematopoiesis, and hepatosplenomegaly. Patients with PV and ET are at an increased risk of developing thrombotic or haemorrhagic events and may progress to an accelerated myelofibrosis phase. All three subtypes are associated with a long-term risk of transformation to acute myeloid leukaemia [Vardiman et al. 2002]. The major presenting symptoms of MPN are related to hypertension or vascular abnormalities, however approximately half of all MPN patients are reported to be asymptomatic at diagnosis [Tefferi and Vardiman, 2008]. PV and ET are diagnosed at a rate of 1 to 3 cases per 100,000 population per year, whereas PMF is rarer [Johansson et al. 2004]. MPN is predominantly diagnosed in the age range of 50–60 years, although it can be diagnosed in younger individuals, particularly if there is a familial predisposition [Bellanne-Chantelot et al. 2006]. A recent epidemiological study found life expectancy of all MPN subtypes is reduced when compared with the general population [Hultcrantz et al. 2012] with PMF having the shortest survival rates [Hoffman and Rondelli, 2007].

Acquired mutations and cytogenetic abnormalities

Acquired somatic changes are well characterized in MPN, where clonal proliferation is driven by an increasingly diverse and somewhat complex set of genetic abnormalities [Cross, 2011]. By far the most recurrent abnormality in MPN is JAK2 V617F, a point mutation that activates the JAK2 cytoplasmic tyrosine kinase, deregulating downstream signalling pathways that drive myelopoiesis. JAK2 V617F is found in approximately 95% of PV cases, and 50 to 60% of ET and PMF [Baxter et al. 2005; James et al. 2005; Kralovics et al. 2005; Levine et al. 2005]. One third of the remaining PV cases carry JAK2 exon 12 mutations, often accompanied by an apparently isolated erythrocytosis [Scott et al. 2007]. Unlike V617F where only a single codon is affected [Dusa et al. 2008], there have been more than 40 different exon 12 mutations identified to date, most being complex insertion/deletion events affecting residues 536–547 [Scott, 2011]. Approximately one third of JAK2 V617F positive PV and PMF cases are homozygous with mutation burdens greater than 50%. This homozygosity generally occurs as a consequence of mitotic recombination of chromosome 9p, and is referred to as acquired uniparental disomy (UPD) or copy number neutral loss of heterozygosity (CNN-LOH) [Baxter et al. 2005; James et al. 2005; Kralovics et al. 2005; Levine et al. 2005]. In contrast, patients with ET have mutation levels predominantly in the heterozygous range of less than 50% [Scott et al. 2006]. Other mutations are acquired in MPN at varying frequencies, but are all less common than JAK2 V617F. These are listed in Table 1, and include MPL [Pikman et al. 2006], CBL [Grand et al. 2009; Sanada et al. 2009], TET2 [Delhommeau et al. 2009; Langemeijer et al. 2009; Tefferi et al. 2009], SH2B3 [Oh et al. 2010; Pardanani et al. 2010a], ASXL1 [Carbuccia et al. 2009], EZH2 [Ernst et al. 2010; Nikoloski et al. 2010], and DNMT3a [Abdel-Wahab et al. 2011; Stegelmann et al. 2011]. There are additional somatically acquired mutations associated with transformation to secondary acute leukaemia or disease progression, and this list of genes includes many that are also associated with de novo AML, such as TP53 [Harutyunyan et al. 2011b], RUNX1 [Ding et al. 2009; Klampfl et al. 2011], NPM1 [Falini et al. 2005], FLT3, NRAS [Bacher et al. 2007], IDH1/2 [Figueroa et al. 2010; Green and Beer, 2010; Tefferi et al. 2010a], IKZF1 [Jager et al. 2010], SF3B1 [Malcovati et al. 2011], and SRSF2 [Zhang et al. 2012a].

Table 1.

Acquired mutations characterized in sporadic MPN.

Gene	Mutation/location	Genomic location	Protein	Frequency of mutation in sporadic MPN, % cases
				PV	ET	PMF	Blast phase
JAK2	V617F, exon 14	9p24	JAK2	95–97	50–60	50–60	50
JAK2	Various indels, exon 12	9p24	JAK2	1–2	rare	rare	NI
MPL	W515K/L/A, S505N, exon 10	1p34	TpoR	Rare	3–5	5–10	NI
CBL	Point mutations, exons 8 and 9	11q23	CBL	Rare	Rare	5–10	NI
TET2	Mutations across gene	4q24	TET2	10–20	5	10–20	20
SH2B3	Various mutations, mainly exon 2	12q24	LNK	1	3–6	3–6	10
ASXL1	Mutations across gene	20q11	ASXL1	2–5	2–5	13–20	20
EZH2	Various mutations across gene	7q36	EZH2	1–3	1	5–10	NI
DNMT3a	Mutations across gene	2p23	DNMT3a	5–10	1–5	5–10	20
IDH1/IDH2	Mainly IDH1 R132 or IDH2 R140	2q23/15q26	IDH1/2	Rare	Rare	6	20–36

ET, essential thrombocythaemia; MPN, myeloproliferative neoplasm; PMF, primary myelofibrosis; PV, polycythaemia vera; NI, not investigated.

A minority of MPNs display karyotypic abnormalities, the most recurrent being gain of part or all of chromosome 9 in PV, and this is associated with a copy number gain of JAK2 V617F. Gain of chromosome 8, partial trisomy for 1q, and interstitial deletions of 13q and 20q have been recorded in all MPN subtypes. Acquired UPD 1p, 4q 7q, 9p, and 11q is usually associated with homozygosity for mutations in MPL, TET2, EZH2, JAK2 and CBL, respectively [Score and Cross, 2012]. However, the picture is incomplete as not all cases with acquired UPD in these regions have one of these corresponding mutations, and there are other, rarer regions of acquired UPD that have been detected in MPN for which mutations remain to be identified.

It is well accepted that chromosomal rearrangements and somatically acquired mutations drive clonal proliferation in MPNs, but much less is known about what factors contribute to the phenotype and severity of disease. The list of recurrent mutations isolated in MPN outlined in Table 1 is set to expand with new sequencing technologies, but bar a few exceptions (JAK2 exon 12 mutations, for example) none of these mutations correlate tightly with haematologically defined disorders. Whilst the identification of phenotype-inducing mutations remains a possibility, it seems unlikely that specific somatically mutated genes or combinations of somatically mutated genes will fully explain the diversity of MPN, for example the reason why some individuals with JAK2 V617F develop PV whereas other develop ET. An alternative possibility is that constitutional genetic variation drives much of the observed phenotypic heterogeneity between individuals.

Genetic predisposition to MPN

As illustrated in Figure 1, there are two main patterns of inherited predisposition to true, clonal MPN: (a) ‘true’ familial MPN with Mendelian inheritance, with the underlying risk alleles presumed to be rare and relatively highly penetrant; and (b) a more general predisposition, driven by weakly penetrant, common risk factors. In addition, mutations in several genes have been associated with hereditary erythrocytosis or thrombocytosis, disorders that are nonclonal which may occasionally be mistaken for true MPN. This review will discuss each of these in turn.

Figure 1.

Inherited predisposition to MPN and related disorders. Germline predisposition factors sit on a sliding scale of frequency and disease penetrance.

Familial MPN

There are many reports in the literature of families with two or more members with MPN, and this clustering is an observation that exceeds that expected by chance in the general population. More detailed analysis of MPN pedigrees indicates the inheritance patterns are markedly heterogeneous, suggesting there are probably a variety of different germline mutations driving the predisposition. For many families, MPN is inherited in an autosomal dominant pattern [Kralovics et al. 2003b; Rumi et al. 2006], whereas for other families penetrance of disease is somewhat decreased, and some have a recessive pattern [Rumi et al. 2007]. It has been suggested that the age of onset is lower in familial MPN compared with sporadic cases [Bellanne-Chantelot et al. 2006; Kralovics et al. 2003b], and there is some evidence for anticipation between generations [Rumi et al. 2007]. As for clinical phenotype, approximately 60% of MPN families are characterized by individuals who all present with the same disease subtype, i.e. typically all PV or all ET whereas the remaining 40% of families have affected members with different diagnoses of MPN. Occasional families are characterized by individuals affected by more diverse myeloid malignancies including chronic myeloid leukaemia (CML) or systemic mastocytosis (SM) [Bellanne-Chantelot et al. 2006]. There appears to be a stronger propensity for family members with PMF to transform to acute leukaemia [Rumi et al. 2007].

Whatever the driving factor, the majority of familial MPN cases have a disease that closely mirrors that of sporadic MPN, with similar clinical features and thrombotic and haemorrhagic complications, and risk of progression to myelofibrosis and leukaemia. Moreover, it has been shown that haematopoietic progenitors from select cases of familial MPN were clonal and were able to form erythropoietin-independent erythroid colonies in vitro, a hallmark of sporadic PV and a feature not observed in healthy individuals or cases with idiopathic erythrocytosis [Kralovics et al. 2003a]. It has been estimated that as many as 5–10% of sporadic MPNs actually have a family history, a measure that has become more easily defined now there are common molecular markers for discriminating MPN from other hereditary conditions. Affected family members with MPN carry mutations associated with sporadic MPN, but the vast majority of these are not inherited. The genetic basis for this familial clustering is not yet known despite several linkage studies attempting to identify candidate regions [Bellanne-Chantelot et al. 2006]. The task of locating abnormalities is likely to have been hampered by the fact that MPNs are late-onset diseases, penetrance appears to be variable and there be a considerable diversity of predisposition loci [Rumi et al. 2007]. Whatever these inherited genetic factors turn out to be, they cannot be the sole cause of clonal proliferation because development of MPN requires the acquisition of additional somatic mutations. It is tempting to hypothesize this factor could contribute to increased genomic instability at the level of DNA replication or DNA repair in response to DNA damaging agents or replication stress of immunologically challenged stem cells, perhaps resulting in a weaker predisposition than that seen with inherited mutations in true tumour suppressor genes or DNA repair proteins [Assumpcao et al. 2008; Meindl et al. 2010].

To date, JAK2 V617F together with MPL W515L/K and inactivating TET2 mutations remain the most recurrently acquired mutations in both sporadic and familial MPN [Saint-Martin et al. 2009]. JAK2 V617F has never been reported to be inherited through the germline but other weaker activating alleles appear to be responsible for occasional families with hereditary thrombocytosis (see below). Similarly, MPL W515L/K is not inherited but other alleles may pass through the germline and again be associated with hereditary thrombocytosis. TET2 mutations are generally acquired, but a single family has been described with an inherited truncating mutation that may have predisposed to true clonal MPN [Schaub et al. 2010].

Common variation and the JAK2 46/1 haplotype

One of the main lines of evidence for the presence of common susceptibility genes in the general population comes from an important Swedish epidemiological study which revealed a fivefold to sevenfold increased relative risk (RR) of developing MPN among first-degree relatives of MPN patients. This risk increases to 12 fold when comparing the risk of ET in first-degree relatives of ET patients [Landgren et al. 2008]. In humans, evidence that inherited single nucleotide polymorphisms (SNPs) contribute to MPN development was first suggested by Pardanani and colleagues, who linked SNPs within the JAK2 and EpoR genes with specific MPN subtypes [Pardanani et al. 2008].

Subsequently the unexpected finding was made that the JAK2 V617F mutation does not arise randomly, but rather preferentially appears on a particular haplotype of JAK2 referred to as ‘46/1’ [Jones et al. 2009] or ‘GGCC’ [Olcaydu et al. 2009a]. This haplotype is seen at a frequency of 24% in European populations, meaning that roughly 50% of individuals carry at least one allele [Jones et al. 2009; Kilpivaara et al. 2009; Olcaydu et al. 2009a]. The single 46/1 locus accounts for an estimated 50% of the population attributable risk of developing an MPN, that is the proportion of the risk of developing MPN conferred by inherited factors. Moreover, JAK2 46/1 is one of the strongest predisposition factors linked to development of a molecularly defined malignancy identified to date with an odds ratio of developing MPN being threefold to fourfold higher in patients carrying a 46/1 allele compared with noncarriers. Although 46/1 is likely to be the most important common risk factor for MPN development, its penetrance is very low and it cannot be used to predict disease development. Furthermore, the JAK2 46/1 haplotype does not explain familial MPNs [Olcaydu et al. 2011] or the phenotypic diversity associated with JAK2 V617F. Recent reports have confirmed the association between 46/1 and MPN in Japanese [Ohyashiki et al. 2012] and Chinese cases [Zhang et al. 2012b], indicating that the underlying mechanism is not limited to Caucasians and must therefore have a relatively old evolutionary basis.

Patients with normal karyotype acute myeloid leukaemia (NK-AML) have an excess of JAK2 46/1, and the data indicated a trend towards shorter survival and myelomonocytic proliferation [Andrikovics et al. 2010; Nahajevszky et al. 2011]. This is an interesting finding especially in the context of a recent population-based investigation into risk of myeloid and lymphoid diseases in first-degree relatives of adults with AML and myelodysplastic syndrome (MDS). For AML patients over the age of 21, the most significant risk in relatives was for PV (RR 2.3), suggesting that shared predisposition alleles predispose more broadly to myeloid malignancy, and that JAK2 46/1 may prove a good candidate locus in this situation [Goldin et al. 2012]. The association of 46/1 with a wider series of haematological disorders suggests the biological consequence(s) of this single predisposition locus could manifest in a wider range haematopoietic cell types and contribute more broadly to the pathogenesis of leukaemia.

Underlying reasons for the association between 46/1 and MPN

It is not clear why acquisition of JAK2 V617F is associated with a particular inherited background. The 46/1 haplotype spans 180–220kb and is made up of hundreds of SNPs in tight linkage disequilibrium (LD). Several SNPs have been used to define or tag 46/1 but these are simply markers and the true causal variant(s) could be anywhere in the LD block and may even be further away. The block contains the whole or most of the JAK2 gene plus INSL4 and INSL6, two genes that are not believed to be expressed in haematopoietic cells. It seems highly likely therefore that the underlying variant(s) in some way influence JAK2, although it is possible that there could be long-range effects on other genes.

Two broad hypotheses have been suggested to explain the observed association. The first, termed ‘hypermutability’ considers 46/1 as more genetically unstable, acquiring V617F at a faster rate than other haplotypes. In the second, called ‘fertile ground’, V617F is suggested to arise on all haplotypes at an equal rate, but there is an additional factor on 46/1 which provides a selective advantage to the V617F-positive clone. Support for the hypermutability hypothesis comes from the observation that V617F has apparently arisen at least twice in some individuals [Lambert et al. 2009; Olcaydu et al. 2009a], and possibly by the fact that JAK2 exon 12 mutations are also associated with 46/1, albeit with a weaker risk [odds ratio (OR) 1.81–2.10] [Jones et al. 2010; Olcaydu et al. 2009b]. There are also reports of isolated cases which carry biclonal JAK2 V617F and exon 12 mutations [Li et al. 2008], a finding that appears to be highly improbable given the low incidence of MPN. Although this might be explained by a genetically unstable JAK2 haplotype, it is also possible that such individuals have an inherited or acquired propensity to acquire further mutations, not only in JAK2 but also in other genes as well. Indeed several cases that are positive for both JAK2 V617F and BCR-ABL or MPL 515 mutations have been reported [Hussein et al. 2007; Lasho et al. 2006] which lends support to the notion of a general hypermutability that is independent of 46/1.

We have argued that the finding that JAK2 46/1 is also associated with JAK2 V617F-negative MPNs (albeit more weakly; OR 1.24) [Jones et al. 2010; Pardanani et al. 2010b; Tefferi et al. 2010b], as well as MPNs that are V617F-negative but carry the MPL W515 mutations (OR 1.44) [Jones et al. 2010], is more in keeping with the idea that there is some functional difference on 46/1 that contributes to disease development. Other, much smaller, studies however have not confirmed the association between 46/1 and MPL mutated MPN and further studies are needed to define their relationship [Pietra et al. 2012; Tefferi et al. 2010b].

If JAK2 on the 46/1 haplotype is functionally different, it might be expected to influence clinical phenotype of MPN. This is currently a matter of debate, with some studies failing to find any link to clinical parameters [Guglielmelli et al. 2010; Jones et al. 2010; Kouroupi et al. 2011; Pardanani et al. 2010b], and others providing interesting correlations with 46/1. One study has shown that the absence of 46/1 is linked to inferior survival in PMF [Tefferi et al. 2010b], but this remains unsubstantiated [Guglielmelli et al. 2010]. It has been reported that in untreated PV patients with homozygous 46/1, the V617F allele burden is more likely to progressively increase over a year when compared to PV cases that did not have any 46/1 alleles, suggesting 46/1 might facilitate greater expansion of V617F homozygous clones [Alvarez-Larran et al. 2012]. A recent study in Chinese MPNs revealed 46/1 is associated with platelet count in PV and PMF, and raised haemoglobin and haematocrit in ET [Wang et al. 2013]. A significant excess of 46/1 alleles has been reported in JAK2 V617F-positive splanchnic vein thrombosis (SVT) patients, and the presence of 46/1 has been correlated with increased erythropoiesis in JAK2 V617F negative cases with SVT [Smalberg et al. 2011]. It might also be expected that 46/1 could influence blood counts in normal individuals. Indeed one study involving approximately 50,000 Danish individuals correlated the presence of 46/1 alone with an elevation in erythrocyte count, and reduction in platelet count [Nielsen et al. 2013]. However no such associations were detected in similarly sized genome-wide association studies (GWASs) [Gieger et al. 2011; Newton-Cheh et al. 2009; van der Harst et al. 2012].

Other support favouring the fertile ground hypothesis comes from the association of 46/1 with haematopoietic colony numbers in normal individuals [Jones et al. 2009] and, more convincingly, a 46/1 tag SNP having robust association with inflammatory disorders, such as Crohn’s disease [Barrett et al. 2008], ulcerative colitis [Anderson et al. 2011a], and psoriasis [Ellinghaus et al. 2012]. Indeed, epidemiological studies have shown MPN patients are at a greater risk of developing a range of autoimmune/auto inflammatory diseases before or after onset of MPN [Barosi et al. 2010; Kristinsson et al. 2010; Rondeau et al. 1983]. GWASs in Crohn’s disease and ulcerative colitis have also revealed significant associations with the IL-23 receptor, IL12B, TYK2, and STAT3, suggesting whole signalling pathways centred around JAK2 could be functionally different and collectively contribute to the pathogenesis of these disorders [Lees et al. 2011]. So far there has been no evidence for other shared risk factors between MPN and autoimmune disorders apart from 46/1. Given the importance of JAK-STAT signalling in MPN, the Crohn’s associated STAT3 variant was investigated in MPN but not found to be significantly different from controls [Jones and Cross, 2010]. More systematic searches are required to determine if there are other shared genetic susceptibility factors between MPNs and immune-related disorders.

Despite these considerations, the biological mechanism behind the association of the 46/1 haplotype and acquisition of JAK2 and other mutations remains uncharacterized. The finding that there were no sequence changes on the 46/1 allele, and no JAK2 expression changes detected in peripheral blood leukocytes suggests that if there is a functional variant, its effects are subtle and perhaps restricted to a particular haematopoietic progenitor cell. A small functional effect would also be consistent with the observation that 46/1 is not required for disease development, as V617F arises on a different haplotype in approximately 25% of JAK2 mutated MPNs. It is also possible that the hypermutability and fertile ground hypotheses are not mutually exclusive and elements of both may be important. Finally, the hypothesis that there is a single causal variant on 46/1 may not be correct and it may turn out that different underlying variants may be driving different associations.

The phenotypic diversity of MPN

One of the most interesting unsolved questions in MPN relates to understanding what factors drive the phenotypic diversity associated with particular mutations. In some instances it appears that particular mutations are strongly associated with clinical phenotype. JAK2 exon 12 mutations have only been detected in cases with V617F negative PV and idiopathic erythrocytosis [Percy et al. 2007b; Scott et al. 2007]. Mutations at JAK2 R683 are seen in a small proportion of children with acute lymphoblastic leukaemia [Bercovich et al. 2008; Kearney et al. 2009], and the presence of JAK2 R683 mutations correlated with aberrant expression of CRLF2, a cytokine receptor that signals through JAK2 [Hertzberg et al. 2010]. Mutations in MPL W515 have only been detected in ET and PMF [Pikman et al. 2006]. Further afield, atypical MPNs such as SM are characterized by mutations in the stem cell factor receptor gene (KIT D816V) [Furitsu et al. 1993], and FIP1L1-PDGFRA is usually restricted to cases with hypereosinophilic syndrome [Cools et al. 2003]. Other gene fusions, such as those involving FGFR1, also show marked phenotypic correlations [Aguiar et al. 1997; Cross and Reiter, 2008]. Very recently, SETBP1 mutations have been associated with atypical CML [Piazza et al. 2012]. Current evidence shows a heterogeneous combination of mutations can be acquired in each MPN patient, and data is emerging to suggest that the order of acquisition of mutation may be important. However, it is likely that inherited factors also play an important role in determining phenotype.

The question why do some JAK2 V617F patients develop PV whereas others develop ET remains unanswered. Homozygosity for V617F is more common in PV compared with ET, leading to the suggestion that increased JAK2 V617F signalling is associated with a more erythrocytic phenotype [Baxter et al. 2005; James et al. 2005; Kralovics et al. 2005; Levine et al. 2005]. This suggestion has some support from transgenic mouse models [Tiedt et al. 2008] and also the fact that JAK2 exon 12 mutations are more strongly activating that V617F. However, the strength of JAK2 signalling cannot be the whole story since many PV patients only carry heterozygous clones [Scott et al. 2006]. Moreover, small homozygous V617F positive clones have been detected at low frequency in a substantial proportion of ET patients, but they appear to be constrained from expanding or they have no selective advantage [Godfrey et al. 2012]. A qualitative difference in the signalling state of STAT proteins has been described in MPN; ET progenitors had high phosphorylation levels of STAT1 and STAT5, whereas PV progenitors had only phosphorylated STAT5 [Chen et al. 2010]. The reasons behind this and other phenotypic differences are unclear, but are potentially the result of a complex interplay between acquired and/or inherited variation, and possibly environmental exposure, all unique to each MPN patient.

Other evidence that the genetic background may influence MPN phenotype comes from mouse models expressing JAK2 V617F, where disease phenotype varies between mouse strains. When the JAK2 V617F transgene was expressed in the BM of C57Bl/6 mice, the animals displayed erythrocytosis with mild leukocytosis. However when JAK2 V617F was expressed in Balb/c mice, a different genetic background, the mice displayed erythrocytosis with dramatically increased leukocyte count and bone marrow fibrosis [Lacout et al. 2006; Wernig et al. 2006; Zaleskas et al. 2006].

Other predisposition factors in MPN

Whilst JAK2 46/1 accounts for a large component of the inherited risk for developing JAK2 V617F positive MPN [Landgren et al. 2008], it explains only a very minor component of JAK2 V617F negative disease, suggesting that there may be other genetic risk variants. Recently, some reports of other common predisposition alleles linked to MPN have begun to emerge. Like JAK2 46/1, these loci were largely discovered using candidate gene approaches rather than systematic screens.

By focusing on genes involved in relevant signalling pathways, an excess of the glucocorticoid receptor β (GRβ) SNP A3669G (rs6198) was recently identified in PV but not ET cases. Expression of higher levels of a dominant negative isoform of GRβ appears to block stimulation by glucocorticoids and contribute to increased erythrocytosis [Varricchio et al. 2011]. The A3669G variant has also been associated with blastic transformation in JAK2 V617F positive PMF (OR of between 1.6 and 1.8), where the presence of the risk allele G/G correlated with a higher leukocyte count at diagnosis, splenomegaly, and higher numbers of circulating CD34+ cells [Poletto et al. 2012]. In another study, common polymorphisms in DNA damage repair pathway genes were investigated for their contribution to transformation of ET and PV to acute leukaemia. In the ERCC2 (XPD) gene the K751Q (rs13181) minor allele was statistically associated with both leukaemic transformation (OR 4.9) and development of nonmyeloid malignancies (OR 4.2). This significance remained even after controlling for the type of treatment that the patients received, as it has been suggested carriers of this risk allele might be more susceptible to the leukaemogenic effects of long-term exposure to particular therapies [Hernandez-Boluda et al. 2012].

Following on from the observation that IDH-mutated glioblastomas are significantly enriched for risk variants in CCDC26, these polymorphisms were screened in a series of acute and chronic myeloid malignancies that also carry IDH mutations. Despite small numbers, modest associations with variants in CCDC26 were observed between cases that carried particular IDH mutations versus IDH-unmutated counterparts [Lasho et al. 2012]. Very recently, a small GWAS identified MPN-predisposition variants in the TERT gene which encodes a telomerase reverse transcriptase, and a weaker association in the ataxia telangiectasia mutated gene, ATM. These associations were independent of JAK2 V617F status, and were made in a collection of different MPNs recruited in a web-based study where patients self-reported their diagnosis as being ET, PV, PMF, SM, or CML [Hinds et al. 2012]. Lastly, an isolated case of PV who developed anaemia and later transformed to acute leukaemia was found to have acquired UPD on chromosome 14q. On further investigation, an inherited heterozygous nonsense mutation in the Fanconi anaemia complementation group M (FANCM) gene, located at 14q, was found to be reduced to homozygosity [Harutyunyan et al. 2011a]. Whether this directly contributed to development of anaemia is unclear but no other candidate mutations were identified.

In an attempt to map out common predisposition factors more systematically, we are currently undertaking a GWAS to identify common SNPs with frequencies that differ significantly between unrelated Caucasian JAK2 V617F negative MPN patients and ethnically matched healthy controls. This is a powerful approach for identifying disease-associated regions with no prior hypothesis, as the SNPs genotyped in both groups tag a large proportion of common genetic variation across the genome. Preliminary results indicate there are no disease associations as strong as that seen for JAK2 46/1, which is perhaps not surprising given the heterogeneous nature of JAK2 V617F negative MPN, but some more modest associations are emerging close to interesting candidate genes.

Finally, there is mounting evidence to suggest inherited factors may have a broader effect on susceptibility to developing malignancy, as epidemiological studies have shown MPN patients are at a greater risk of developing both haematological and nonhaematological cancers. MPN patients have a 2.8- to 5.5-fold increased risk of developing a lymphoid neoplasm [Rumi et al. 2011; Vannucchi et al. 2009]. Patients with ET, PV and CML are at slightly increased odds of developing nonhaematological cancers, in the range of 1.2- to 1.6-fold [Frederiksen et al. 2011]. These associations could in part be accounted for by prior leukaemogenic treatments which promote further malignancy, or for other haematological malignancies possibly the presence of V617F itself, which has been linked to increased risk of secondary cancer and DNA damage [Nielsen et al. 2011; Plo et al. 2008]. It has been reported that there is a modest increased incidence of PV in people of Jewish ancestry (5.8% versus 3%) in the French population [Najean et al. 1998]. More strikingly, the incidence of MPN was tenfold higher in Ashkenazi Jews compared with other populations in Northern Israel [Chaiter et al. 1992], suggesting MPN predisposition alleles might cluster in particular groups of individuals.

Hereditary erythrocytosis and thrombocytosis

In other hereditary myeloproliferative disorders, the causative mutation typically segregates with all affected family members, and only one lineage predominates in a polyclonal fashion. Being nonneoplastic, affected individuals are highly unlikely to progress to more aggressive disease. Most often referred to as hereditary erythrocytosis [Gordeuk et al. 2005], or hereditary thrombocytosis, several causative mutations have been well characterized and either target important components of cytokine signalling pathways or components of oxygen sensing pathways. These disorders have been classified into four groups (ECYT1-4) by Online Mendelian Inheritance in Man (OMIM), and have been reviewed extensively elsewhere [McMullin, 2010; Skoda, 2010], but we include a brief description here as occasional cases may be confused with true MPN.

Erythropoietin receptor (EPOR) mutations have been described in cases of hereditary erythrocytosis, primary familial and congenital PVs (PFCPs). Here, erythrocytosis is driven by a truncated EPOR protein, rendering progenitor cells hypersensitive to stimulation by erythropoietin (EPO), which is present at low levels in circulating serum. All EPOR mutations identified to date truncate the cytoplasmic domain of the EPOR protein, resulting in loss of a negative regulatory region at the C-terminus. PFCP patients are more likely to suffer from thrombotic and haemorrhagic events, and premature morbidity and mortality have been reported [Al-Sheikh et al. 2008; de la Chapelle et al. 1993; Prchal, 2005; Prchal et al. 1985]. Approximately 10–20% of all PFCP cases carry a truncating EPOR mutation, but the molecular basis for the remaining families remains unclear, despite extensive genetic investigations focusing on components the EPO pathway and regulation of EPO expression. Constitutional mutations in genes encoding components of the oxygen sensing pathways have been characterized in a small minority of familial polycythaemias with inappropriately normal or high circulating EPO levels. The best characterized are autosomal recessive mutations in the von Hippel–Lindau (VHL) gene, which impair the rate of ubiquitin-mediated degradation of the transcription factor hypoxia-induced factor (HIF)-1α. Consequently, raised HIF-1α heterodimers upregulate expression of EPO and other genes, stimulating erythrocytosis and causing other pathological consequences such as thrombocytosis and cerebrovascular events [Ang et al. 2002]. VHL mutations were first characterized in patients with hereditary erythrocytosis from the Chuvash Republic in central Asia where the homozygous VHL (598C>T) mutation is endemic [Ang et al. 2002; Sergeyeva et al. 1997] but there are different VHL mutations that have been detected in pedigrees from different ethnic groups [Bento et al. 2005; Cario et al. 2005; Pastore et al. 2003]. Patients with homozygous VHL mutations do not develop tumours that are characteristic of the classical autosomal dominant VHL syndrome with a high incidence of malignant and benign neoplasms of the central nervous system and kidneys [Kim and Kaelin, 2004]. In addition, constitutional missense mutations of the prolyl hydroxylase domain protein 2 (PHD2) have been reported, also causing erythrocytosis in a small number of families [Albiero et al. 2012; Kim and Kaelin, 2004; Ladroue et al. 2012; Percy et al. 2006, 2007a]. PHD2 is critical for mounting a correct response to hypoxia, as it mediates oxygen-dependent hydroxylation of HIF1α on proline residues, facilitating binding of VHL to HIF1α, and necessary proteasomal degradation. HIF2α may also be targeted by missense mutations in familial erythrocytosis [Gale et al. 2008; Martini et al. 2008; Percy et al. 2008a, 2008b]. Functional studies indicate HIF2α mutations impair both PHD-2-induced hydroxylation of HIF2α, and the subsequent recognition of the hydroxylated HIF2α by VHL, leading to enhanced stabilisation of HIF2α and transcriptional upregulation of EPO [Furlow et al. 2009].

Hereditary thrombocythaemia (also called familial thrombocytosis) has been linked to defects in the thromobopoietin signalling pathway, particularly mutations that target thromobopoietin (TPO) or its receptor (MPL). Affected individuals have elevated circulating levels of TPO, which aberrantly stimulates megakaryopoiesis and cause excessive platelet production, without any involvement of other lineages. It has long been known that for some hereditary thrombocytosis cases, elevated TPO levels are a consequence of increased translational efficiency of TPO, caused either by a G to C mutation [Liu et al. 2008; Wiestner et al. 1998] or a one-base-pair deletion [Ghilardi and Skoda, 1999; Ghilardi et al. 1999; Kondo et al. 1998] affecting the splice donor site of intron 3. These mutations lead to TPO mRNAs with shortened 5’ untranslated regions that are more efficiently translated into mature protein than normal transcripts. Hereditary thrombocythaemia cases typically have an earlier onset of disease than those with sporadic ET, but share similar frequencies of thrombosis and haemorrhagic episodes.

There are also rare examples of activating MPL mutations passed through the germline, all associated with marked thrombocytosis and other myeloproliferative features, including splenomegaly, bone marrow fibrosis and higher risk of thrombosis. Some of these MPL mutations activate signalling pathways in vitro, and include amino acid changes S505N [Ding et al. 2004; Teofili et al. 2010], P106L [El-Harith et al. 2009], and W515R [Vilaine et al. 2012]. Interestingly, MPL K39N is a functional polymorphism restricted to the African American population, whereby the presence of one risk allele was reported to correlate with elevated platelets in the peripheral blood and presence of two copies was linked to severe thrombocytosis [Moliterno et al. 2004]. Finally, there are isolated examples where inherited JAK2 mutations segregate with features of disease; JAK2 mutations V617I, R564Q, and H608N are associated with a phenotype characterized by thrombocytosis and high penetrance [Etheridge et al. 2011; Mead et al. 2012; Rumi et al. 2012]. It appears that these mutations are more weakly activating that V617F, which may explain their ability to be passed through the germline.

Conclusions

Knowledge about the genetic basis of sporadic MPN has increased enormously since the discovery of the JAK2 V617F mutation, and this pace of progress is set to continue as new sequencing technologies further characterize new somatically acquired mutations. Furthermore, the characterization of JAK2 46/1 as a strong, common risk factor for the acquisition of V617F mutations was an important finding in terms of understanding genetic predisposition to MPN, hinted at by family and epidemiological studies. It is clear that germline factors play an important role in the pathogenesis of MPN, and it will be informative to see how this inherited variation modifies initiation, phenotype, clinical course and transformation driven by acquired mutations (Figure 2). Further candidate and systematic studies will define more risk variants with variable frequencies and degrees of penetrance, but none are expected to be as striking as that observed for JAK2 46/1. Far less is known about what drives familial MPN, but it is expected next-generation sequencing will enable the discovery of further rare, highly penetrant variants. Finally, environmental and lifestyle influences on susceptibility to MPN have been investigated and reviewed elsewhere [Anderson et al. 2011b]. These studies have mainly focused on exposures to various chemicals through lifestyle choices or occupation, and despite the vast heterogeneity in defining such hard to measure risks, there are recurrent reports linking exposure to benzene to the development of PV [Aksoy, 1980; Bernardini et al. 2005]. An ultimate goal would be to try to understand how rare and common germline factors, somatically acquired mutations and environmental modifiers all contribute to the development and clinical course of MPN.

Figure 2.

Inherited factors predispose to the development and influence the clinical course of MPN.