Molecular Classification of Myeloproliferative Neoplasms—Pros and Cons
- First Online:
- Cite this article as:
- Qureshi, M. & Harrison, C. Curr Hematol Malig Rep (2013) 8: 342. doi:10.1007/s11899-013-0179-9
- 380 Views
Dameshek first postulated a common myeloproliferative heritage for the myeloproliferative disorders, now termed neoplasms. This prescient observation was validated by the description of a common mutation in exon 14 of JAK2 for patients with essential thrombocythemia, polycythemia vera and primary myelofibrosis. In recent years, our knowledge of the molecular abnormalities underpinning these disorders has expanded significantly. At the same time, we have continued to use a classification based largely upon the first clinical descriptions of these entities, which sometimes proves problematic in differentiating between these conditions and normal reactive processes, myelodysplasia and between the myeloproliferative neoplasm entities themselves. Here, we discuss the pros and cons of a molecular classification and its potential utility in diagnosis, prognosis, and therapeutics.
KeywordsMyeloproliferative neoplasmsMolecular pathologyPolycythemia veraEssential thrombocythemiaPrimary myelofibrosisJAK/STATEpigeneticsHematologic malignancy
Over the past two centuries, our understanding of myeloproliferative neoplasms (MPNs) has evolved from clinical and haematopathological observations to an increasing appreciation of the molecular mechanisms underpinning the neoplastic process and their interplay with clinical phenotype and therapy. The disorders chronic myeloid leukaemia (CML), polycythemia rubra vera (PV), and primary myelofibrosis (PMF) were all identified as clinical and pathological entities in the 19th century. Essential thrombocythemia (ET) was delineated later by Epstein and Goedel in 1934, post-PV MF was described by Hirsch in 1935, and so during the first half of the 20th century, the inter-relationship between these disorders began to be defined . Dameshek, however, was the first author to formally articulate the idea of a common ‘myeloproliferative’ heritage in his landmark publication of 1951: “It is possible that… ‘myeloproliferative disorders’—are all…variable manifestations of proliferative activity of the bone marrow cells, perhaps due to a hitherto undiscovered stimulus” . Dameshek thus importantly postulated not only the possibility of transition between these disorders, but more interestingly, a common primary mechanism for all MPNs. Dameshek’s proposal began to bear fruition during the 1980s with growing evidence that tyrosine kinase (TK) activity provided the molecular mechanism for CML, which is not formally discussed further in this chapter.
Genomic instability has emerged as a fundamental neoplastic process that underlies the hallmarks of cancer as formulated by Hanahan and Weinberg . The underlying molecular genetics of PV, ET and PMF remained enigmatic for many of the years following their original description, but the role of TK proteins in the pathogenesis of myeloproliferation was increasingly scrutinised. These investigations bore fruit in 2005 when the Janus Kinase (JAK)2 V617F TK mutation was identified as a common theme (vide infra). By then, imatinib had demonstrated the potential impact of targeted molecular therapy on TK-mediated neoplasms, and the discovery of the JAK2 mutation consequently generated much excitement. Subsequently, this excitement has become tempered by the realisation that JAK2’s role in MPN is not as homogenous and strictly causational as BCR-ABL’s role in CML. Furthermore, important questions remain regarding the precise role of JAK2 in MPN’s oncogenic pathway, alternative molecular pathways that come into play, the driver mutations for JAK2-negative MPN, epigenetic effects, the explanation of phenotypic diversity within JAK2-positive MPN, and the prognostic relevance of mutational status.
Current Classification of MPN Pros and Cons
An important question that arises from our expanded knowledge of the underlying molecular pathogenesis of MPN is whether we should abandon the current classification of such disorders and apply a molecular classification instead. This question reflects a rational scientific response to novel data, similar to the modernised disease classification of Acute Myeloid Leukaemia (AML). It also reflects difficulties in applying the current World Health Organisation (WHO) classification.
Key features of a successful and meaningful classification are clinical utility, and that it should include features which are reproducible, reliable, and above all, relevant. It should define entities with distinctive behaviour either in natural history or in terms of response to therapy. In addition, the classification must ideally be readily utilised in all the clinical contexts where the disorders are diagnosed or treated.
World Health Organisation Classification for Myeloproliferative Neoplasms (MPN) 2008
Named molecular factor
Chronic myelogenous leukemia
JAK2 V617F / exon 12 or other clonal marker
JAK2 V617F or other clonal marker
JAK2 V617F or other clonal marker
Chronic neutrophilic leukemia
Chronic eosinophilic leukemia, not otherwise specified
FIP1 L1 PDGFR alpha
Mast cell disease
What Might the Key Components of a Molecular Classification Be?
Furthermore, JAK2 can cause transcriptional activation independently of STAT, by directly entering the nucleus and phosphorylating histone 3 on tyrosine 41 (H3Y41), and this impacts the transcriptional repressor heterochromatin protein 1 alpha (HP1α) and regulates Nuclear Factor Erythroid (NF-E2) expression . This pathway provides an example of how JAK2 can directly cause epigenetic effects, and suggests novel therapeutic approaches such as JMJD-demethylase inhibition . Lastly, JAK2 is itself a client protein of its heat shock protein 90 (HSP90) molecular chaperone, and this dependence on HSP90 provides an example of how JAK2 may be targeted through its conformational process rather than its kinase activity. For this reason, cells that are resistant to JAK2 kinase inhibition continue to respond to HSP90 inhibitors that degrade JAK2 [11•].
Constitutive activation of JAK2 via the V617F mutation represents an oversimplification of the complex molecular interactions that regulate the JAK/STAT pathway. For example, other JAK2 mutations such as those within exon 12 can produce constitutive activation. Similarly, MPL mutations within exon 10 can also map to JAK2 signalling pathways. In addition, the JAK/STAT pathway is adaptable by heterodimerization between JAK2 and JAK1 or TYK2, which can bypass JAK2 inhibitory agents [12•]. Lastly, JAK2 V617F mutation is not a homogenous phenomenon, and evidence is emerging of its dynamic relationship with JAK2 wild-type. For instance, Akada et al. have found that hemizygous mice develop a more severe myeloproliferative phenotype when compared to mice that are homozygous for the JAK2 V617F mutation, suggesting that the wild-type allele regulates the mutated allele .
The prevalence of the JAK2 V617F molecular mutation in MPN presents an obvious conundrum in any proposed molecular classification: how can the same mutation produce the phenotypic diversity presented by PV, ET and PMF, if indeed we maintain these separate entities? This problem is addressed to some extent by gene dosage. Transgenic mouse models have demonstrated that low expression of JAK2 V617F produces an ET phenotype, whereas higher expression results in a PV phenotype . Correspondingly, JAK2 V617F homozygosity features commonly in patients with PV rather than those with ET . Additional nuance is provided by the observation that PV is characterised by expansion of a large dominant homozygous V617F subclone, whereas ET subclones are small [16•]. More recently, induced Pluripotent Stem cells (iPS) derived from patients with MPN confirm that JAK2 V617F heterozygous iPS are hypersensitive to thrombopoietin (TPO) and homozygous iPS are independent of TPO . Furthermore, evidence is now emerging that JAK2 V617F homozygosity may produce an ET phenotype in females in preference to a PV phenotype . Besides gene dosage, the mutation itself can produce phenotypic variation. For instance, mouse models of JAK2 V617F demonstrate erythrocytosis and granulocytosis, whereas JAK2 K539I in exon 12 gives erythrocytosis only . This correlates clinically with the observation that exon 12 mutations have been found in PV only, whereas V617F produces more variation .
MPL mutations provide a further entity for inclusion in a molecular classification. These mutations act upstream of JAK2 mutations, as mutated MPL activates the JAK/STAT pathway independently of its TPO ligand. MPL provides an important reminder that dysregulation of the JAK/STAT pathway can produce MPN irrespective of JAK2 mutational status.
Epigenetics and MPN
The relationship between MPN mutations and the epigene is of obvious interest, particularly because several gene mutations in MPN are known to promote epigenetic dysregulation. These include mutations in ten-elevent-translocation-2 (TET2), additional sex combs like 1 (ASXL1) and DNA methyltransferase 3A (DNMT3A).
Regarding the methylation profiles of MPNs, experimental data suggests that PV, ET and PMF have congruous profiles that differ substantively from controls. Transformation to AML is associated with a distinct methylation profile, particularly the Interferon (IFN) pathway . In one study, Global DNA methylation analysis by unsupervised clustering indicated that patients with PV and ET were epigenetically closer to normal controls, whereas five out of a total 12 patients with PMF formed a cluster [22•]. In the same study, subsequent supervised clustering showed hypermethylation in PV and ET samples, which affected several transcription factors including GATA1; by contrast, PMF demonstrated both hypermethylation and hypomethylation. Hypermethylated genes influenced pathways of inflammatory mediators, whereas hypomethylated genes governed cell production, correlating with the clinical complexity of PMF and confirming the idea that neoplasia is a phenomenon involving inflammation and increasingly dysplastic proliferation. No epigenetic clustering was associated with the JAK2 mutation per se.
Nevertheless, the JAK2 mutation does interact with the epigenetic machinery of MPN, for instance by H3Y41, as described above. Similarly, mutated JAK2 V617F phosphorylates protein arginine methyltransferase 5, resulting in reduced methylation of histones H2A and H4 and subsequent epigenetic dysregulation.
The ASXL1 mutation also participates in the epigenetic machinery of MPN, but its precise function remains elusive The wild-type gene encodes a protein which binds chromatin, and locally can both upregulate and downregulate transcription. Mutated or deleted ASXL1 is more frequent in PMF than in PV or ET, and indeed it demonstrates a distinct methylation profile on unsupervised clustering [22•], suggesting epigenetic disruption. TET2 encodes an enzyme that converts 5-methylcytosine to 5-hydroxymethylcytosine, causing DNA demethylation. In murine models, deletion of TET2 results in a myeloproliferative phenotype . There is thus growing evidence that TET2 plays a role in MPN through epigenetic pathways. Many other mutations affecting the epigenetic machinery have been described in MPN. However many of these, in particular TET2 and ASXL1, are not specific for MPN and are found in other myeloid malignancies.
MiRNA signatures are important candidates for future molecular classification of MPN [24•]. MiRNA are short nucleotide non-protein-coding RNAs that bind target mRNA to regulate transcriptional and translational gene expression. Evidence is emerging that miRNA profiles differ between PV, ET and PMF, and that these profiles are independent of JAK2 V617F [25•]. Furthermore, miRNAs act in the stem cell compartment, and therefore could potentially mediate eradication of MPN progenitor cells. Telomerase-mediated immortalisation of neoplastic cells may also emerge as a significant mechanism for MPN, and indeed recent research indicates that a germline variant in the telomerase reverse transcriptase (TERT) gene is associated with PV, ET and PMF, independently of JAK2 V617F mutation status .
What About Prognosis: How is this Best Predicted?
Current molecular mutations with relevance in PMF
Putative prognostic value
The clinical relevance of JAK2 V617F allele burden was demonstrated in a large study by Vannucchi et al., who found that thrombotic events were more common in ET patients with homozygous V617F mutation . Furthermore, gene dosage in both PV and ET was associated with myelofibrotic transformation. Early studies in PMF have given conflicting data, but more recent studies have indicated that low allele burden in PMF is associated with shortened survival and poor prognosis .
Besides V617F, other molecular mutations are emerging as prognostic markers, and Vannucchi et al. have now published data supporting the addition of these markers for PMF [30••]. Two cohorts were studied, and univariate analysis suggested that shortened survival was associated with EZH2, ASXL1 and SRSF2 mutations, and that increased leukaemic transformation was associated with SRSF2 and IDH1 mutations. ASXL1 mutations remained significant on multivariate analysis in the context of IPSS and DIPPS. Interestingly, these mutations are epigenetically florid. Furthermore, they often arise in the context of normal karyotype, and therefore their addition to a new prognostic system may impact on management. This data was also supported by data from the Mayo group [31••].
The group of Spivak has also proposed molecular prognostic scoring of PV, using gene expression profiling to establish a gene panel that predicts aggressive versus indolent disease . Notwithstanding the generally indolent nature of PV, molecular classification may prove useful in identifying patients with high-risk disease who are candidates for aggressive therapy. To date, other than JAK2V617F allele burden, molecular data has not added to prognosis of patients with ET.
Molecular classification of MPN heralds the development of novel therapeutics that provide effective management. These include JAK2 inhibitors, alternative small molecule inhibitors, epigenetic therapies and interferon. Might any of these therapeutic approaches benefit from a molecular classification of MPN?
Ruxolitinib is a novel therapeutic that acts as a potent ATP-competitive inhibitor of both JAK1 (IC50 = 3.3 nmol/L) and JAK2 (IC50 = 2.5 nmol/L) in vitro and decreases IL-6 and TNF-α in murine models . Two phase III studies have demonstrated that Ruxolitinib impacts positively on spleen size, symptoms and circulating levels of IL-6 and TNF- α in humans with PMF [34••, 35••]. It appears to provide these benefits for both JAK2V617F positive and negative patients. Importantly, both studies also showed that best alternative therapies were largely ineffective, and therefore Ruxolitinib has provided the first evidence-supported therapy for PMF. Perhaps more importantly, the persistence of the MPN malignant clone after treatment with Ruxolitinib indicates that JAK2 inhibition in MPN is not equivalent to Bcr-Abl1 inhibition in CML. There is no significant reduction in the JAK2V617F allele burden by Ruxolitinib , nor is there sustained response after cessation of therapy, and whilst there is emerging definitive evidence of long-term survival benefit, this does not seem to relate directly to mutant allele effect.
Several alternative JAK2 inhibitors are currently undergoing trials. SAR302503 is a highly specific JAK2 inhibitor (IC50 = 3 nmol/L) that has recently completed a phase III trial for the treatment of high-risk or intermediate-2 risk PMF in the JAKARTA study. It also inhibits FLT-3, which is a well-established molecular mutation in AML, but of uncertain impact for MPN. In contrast with Ruxolitinib, SAR302503 (formerly TG101348) has demonstrated a significant reduction in JAK2 mutant allele burden in a phase I/II study . Interestingly, it does not appreciably decrease circulating levels of IL-6 and TNF- α, but it does give significant symptomatic relief, thus challenging the hypothesis that MPN symptoms are attributable to a ‘cytokine storm’. Recent evidence presented from a phase II study at the American Society of Hematology (ASH) conference 2012 also demonstrates downstream effects with a reduction in pSTAT3 . Pacritinib is another selective JAK2 and FLT-3 inhibitor, which holds promise for PMF patients with impaired marrow function, as it gives no grade 3 or 4 thrombocytopenia. Furthermore, it is well tolerated, and 11 out of 34 patients with PMF in a phase II study had a significant reduction in spleen volume . Momelotinib (CYT387) is a fourth agent with a specificity similar to that of Ruxolitinib, yet this agent appears to have a different spectrum of activity and adverse events; for example, improvement in anaemia and some adverse events related to grade 1 peripheral neuropathy .
Considerable effort has been made to investigate the mechanisms of MPN resistance to JAK2 inhibitors. Initially, a saturation mutagenesis screen suggested five mutations in the JAK2 kinase domain that conferred JAK2 resistance , but these five mutations have not been identified in patients. An alternative mechanism for resistance, which has been identified in both murine models and patients treated with JAK2 inhibitors, is heterodimerization with other JAK kinases such as JAK1 and TYK2 . Increased understanding of the molecular mechanisms of resistance may lead to more effective therapies.
Inhibitors of Alternative Pathways and Utilising Synergy
Importantly, JAK2 is neither the driver mutation  nor the final effector of MPN, but rather a mediator of STAT activation. A study by Barrio et al. utilised Western blot analysis to demonstrate that a RAF inhibitor (Sorafinib) inhibited ERK, P38 and STAT5, and that a HSP70 inhibitor (KNK437) decreased expression of JAK2 . These inhibitors converge on STAT5, which is proposed as the final effector for MPN. Interestingly, STAT5 knockout mice cannot develop PV, although they can develop PMF . Barrio et al. found that Sorafinib and KNK437 both acted synergistically with Ruxolitinib against HEL and BA/F3 cell lines.
Several further combinational regimens based on molecular classification of MPN have been investigated. Choong et al. have identified synergism of JAK2 and PI3K inhibitors against both cell line and mouse models of MPN . Similar results are given by combination of JAK2 and mTOR inhibitors . Everolimus is an allosteric mTOR inhibitor that is Food and Drug Administration (FDA)-approved for the treatment of several neoplastic disorders. Clinical response against MPN has been demonstrated in a phase II study, but reduction in mutant JAK2 allele burden remains elusive . Bogani et al. have characterised Everolimus’s effects on murine and human JAK2 V617F mutated cell lines as cytostatic, but found that the ATP-mimetic mTOR inhibitor PP242, which inhibits both mTORC1 and mTORC2, is apoptotic . BEZ235 is a dual PI3K/mTOR inhibitor that acts synergistically with SAR302503 and is effective against cells resistant to JAK2 inhibition . Theoretically, dual inhibition should countercheck mTOR inhibition-mediated positive feedback on PI3K and thereby effect more potent suppression of PI3K/AKT/mTOR signalling. Recently, an in vitro study has demonstrated that BEZ235 induces apoptosis of MPN cells . Dual inhibition therefore warrants in vivo studies, both in combination with JAK2 TKI and in subjects who are resistant to TK inhibition.
Combination therapy with HSP90 inhibitors is an alternative strategy to reduce heterodimerization-mediated resistance to JAK2-inhibitors . HSP90 inhibition reduces expression of JAK2 and indeed degrades JAK2. Importantly, HSP90 inhibition provides a treatment strategy for genetic resistance to JAK2 inhibitors [11•, 12•]. The HSP90 inhibitor PU-H71 in conjunction with Ruxolitinib resulted in decreased levels of phosphorylated JAK2, STAT3/5 and AKT in murine studies . HSP90 inhibition again warrants further evaluation in clinical trials.
Epigenetic modification has been tested by two distinct drug strategies. Firstly, histone deacetylase (HDAC) inhibition by Panobinostat, Vorinostat and Givinostat is emerging as a viable treatment paradigm for MPN. Secondly, DNA methyl transferase (DNMT) inhibition by Decitabine provides an alternative route of epigenetic modification.
Panobinostat is a pan-HDAC inhibitor that acts against class II, II and IV HDACs and attenuates JAK2 V617F levels by inhibiting the chaperone function of HSP90 and autophosphorylation of mutated JAK2 . Molecular effects on STAT 3/5 and JAK2 V617F allele burden were encouraging in a phase II trial for the treatment of PMF, but therapy was poorly tolerated . However, lower dosing strategy resulted in improved tolerance and intriguing results for investigators from Mount Sinai where 2/13 patients had a histological response resembling complete remission (CR) . Studies of panobinostat in combination with Ruxolitinib are currently underway .
In JAK2 V617F-expressing HEL cell lines, Vorinostat inhibited STAT3 and STAT5, and increased SOCS1 and SOCS3; this translated to mouse models . Vorinostat decreased JAK2 mutant allele burden in a recent phase II study, but again tolerance was problematic .
HDAC inhibitors may prove to be more tolerable and effective in combination regimens. Givinostat in combination with Hydroxycarbamide (HC) has been tested in a multicentre open-label phase II study of 44 patients with PV who were unresponsive to maximum tolerated doses of HC . This combination was well tolerated and produced complete or partial response in half of patients by European LeukaemiaNet criteria.
Recent work on the HDAC inhibitor Sodium Butyrate sheds light on its molecular mechanism of action. Gao et al. found that Sodium Butyrate hyperacetylates SOCS1 and SOCS3-associated histones H3 and H4 . Cell growth was inhibited in K562 and HEL cell lines ectopically transfected with SOCS1 and SOCS3, indicating that SOCS suppresses oncogenesis. In the same study, selective knockdown of specific HDACs demonstrated that HDAC8 played a significant role in the process of SOCS-mediated inhibition of JAK/STAT signalling.
Demethylating agents such as Azacitidine are already established treatments for Myelodysplastic Syndrome (MDS) and Chronic Myelomonocytic Leukaemia. Cell lines with JAK2V617F mutation and ASXL1 deletion demonstrated genome-wide methylation as compared to normal controls in a recent study by Nischal et al. Decitabine is a cytidine nucleoside analogue and hypomethylating agent, analogous to Azacitidine, and inhibited growth of these cell lines [22•]. However, demethylating therapies have yielded disappointing results in MPN to date .
Interferon α (IFNα) provides enigmatic opportunities to unravel the persistent challenge of MPN: how to eradicate long-term haematopoietic stem cells (LT-HSCs). LT-HSCs are quiescent, they can transplant MPN, they perpetuate disease, and they are not susceptible to JAK2 inhibitors. IFNα activates the JAK/STAT pathway via interferon’s own cognate cytokine receptor. The effects of this upstream activation on LT-HSCs have recently been studied by Mullally et al., who injected purified HSC-enriched cells from both JAK2-mutated and wild-type mice into lethally irradiated recipient mice to produce a chimeric mouse model . Flow cytometry determined that JAK2 V617F disease-propagating stem cells in these mice were depleted by prolonged treatment with IFNα. Furthermore, IFNα resulted in increased cycling LT-HSCs and decreased quiescent cells. The same study found that transplantation of MPN cells from the chimeric mice to congenic recipient mice resulted in impaired engraftment for up to 16 weeks after IFNα, as opposed to 4 weeks without IFNα. This parallels prior work in patients suggesting interferon can induce a molecular remission for JAK2V617F for some patients that may persist for many months ; intriguingly, however, TET2 mutant clones can persist after this therapy.
Future Molecular Strategies
Lastly, molecular classification suggests therapeutic hypotheses that remain to be developed from the bench to the bedside. Cytokine receptors represent one such therapeutic target, because they represent the point of primary activation of the JAK/STAT cascade. Such a strategy is immediately faced with the problem that inhibition of cytokine receptors EpoR and thrombopoietin receptor (TpoR) would retard physiological erythropoiesis and thrombopoiesis. In this regard, cytokine receptors do not represent a homogenous biological mechanism. Indeed, it is known that up to 10 % of patients who have JAK2 V617F negative ET and MF have a myeloproliferative phenotype due to W515 mutations in the TpoR, which permit self-activation . Therefore theoretically, it should be possible to target W515 mutations in the TpoR itself without interfering with the wild-type receptor. Furthermore, in JAK2 V617F positive MPN, it may be possible to target the interaction of both wild-type EpoR and TpoR with JAK2 V617F. Much research remains to be done into the structure of the cytokine receptors, but the success of IFNα suggests that cytokine pathways may eventually participate in the molecular classification of MPN. Telomerase as a potential target was explored by Baerchloter et al. and presented at ASH 2012 . In their study with imetelstat, this agent delivered rapid control of blood counts and also intriguingly rapid and more profound molecular responses than seen with other agents. Inhibition of STAT transcription factors provides another potential treatment strategy that should be less deleterious to normal physiology than inhibition of cytokine receptors, because it acts downstream in the JAK/STAT pathway. Recently, Nelson et al. screened numerous drug therapies for STAT inhibitory effect by using a series of cell lines with luciferase reporter genes specific to single transcription factors . This novel approach established that Nifuroxazide (an antibiotic) inhibits STAT 3, and Pimozide (an antipsychotic) inhibits STAT 5. Nelson et al. found that Pimozide inhibits CML cell lines, even in cells with the T315I mutation, which is resistant to TK inhibitors . The effect of Pimozide on MPN cell lines and murine models remains to be explored.
Disadvantages of Molecular Classification
There are, despite the exciting advances in this field, some obvious and problematic barriers to using solely a molecular classification of the MPNs. There is no single molecular aberration, nor is there a key genetic signature for the MPNs. Many patients with ET and PMF indeed lack the JAK2 V617F mutation, and aside from exon 14 MPL and exon 12 JAK2, the remaining mutations are not unique to MPN. Many of the other mutations, e.g. TET2, are technically much more demanding to detect than JAK2 V617F, since they are spread throughout the gene, which will be a barrier to their widespread introduction as part of a molecular classification. There is emerging and strengthening data linking prognosis with a molecular signature, in particular for PMF; however, this has not yet transferred to the other entities or standard practice, since it requires validation. Lastly, though several new therapies have been identified as a result of molecular understanding, again specific molecular correlates of responders are lacking.
The term MPN incorporates a wide umbrella of disease entities. In this chapter, we have focused on the entities ET, PV ad PMF. The original descriptions of these entities occurred in the late 19th and early 20th centuries. Diagnostic criteria were first proposed by the polycythemia vera study group (PVSG) and later by the WHO, which last modified these criteria in 2008. Currently, a wider and more detailed picture is emerging for the molecular pathogenesis of these diseases, but the incorporation of these findings into risk stratification and therapeutic decisions lags behind. It is the view of the authors that while the 2008 version of the WHO classification is widely utilised, it could be augmented in the future with additional molecular features, but not yet replaced.
Compliance with Ethics Guidelines
Conflict of Interest
Moosa Qureshi and Claire Harrison declare that they have no conflict of interest.
Human and Animal Rights and Informed Consent
This article does not contain any studies with human or animal subjects performed by any of the authors.