Insights into the role of JAK2-I724T variant in myeloproliferative neoplasms from a unique cohort of New Zealand patients

ABSTRACT

Objectives:

This study aimed to compile bioinformatic and experimental information for JAK2 missense variants previously reported in myeloproliferative neoplasms (MPN) and determine if germline JAK2-I724T, recently found to be common in New Zealand Polynesians, associates with MPN.

Methods:

For all JAK2 variants found in the literature, gnomAD_exome allele frequencies were extracted and REVEL scores were calculated using the dbNSFP database. We investigated the prevalence of JAK2-I724T in a cohort of 111 New Zealand MPN patients using a TaqMan assay, examined its allelic co-occurrence with JAK2-V617F using Oxford Nanopore sequencing, and modelled the impact of I724T on JAK2 using I-Mutant and ChimeraX software.

Results:

Several non-V617F JAK2 variants previously reported in MPN had REVEL scores greater than 0.5, suggesting pathogenicity. JAK2-I724T (REVEL score 0.753) was more common in New Zealand Polynesian MPN patients (n = 2/27; 7.4%) than in other New Zealand patients (n = 0/84; 0%) but less common than expected for healthy Polynesians (n = 56/377; 14.9%). Patients carrying I724T (n = 2), one with polycythaemia vera and one with essential thrombocythaemia, had high-risk MPN. Both patients with JAK2-I724T were also positive for JAK2-V617F, found on the same allele as I724T, as well as separately. In silico modelling did not identify noticeable structural changes that would give JAK2-I724T a gain-of-function.

Conclusion:

Several non-canonical JAK2 variants with high REVEL scores have been reported in MPN, highlighting the need to further understand their relationship with disease. The JAK2-I724T variant does not drive MPN, but additional investigations are required to exclude any potential modulatory effect on the MPN phenotype.

KEYWORDS:

• myeloproliferative neoplasms (MPN)
• polycythemia vera (PV)
• essential thrombocythemia (ET)
• primary myelofibrosis (PMF)
• Janus kinase 2 (JAK2)
• JAK2-I724T
• New Zealand Polynesians
• Māori

Introduction

Classical Philadelphia-chromosome-negative myeloproliferative neoplasms (MPN) are characterized by specific driver mutations affecting Janus kinase 2 (JAK2), calreticulin (CALR), or myeloproliferative leukemia (MPL) genes [1]. These mutations are present in 85–90% of MPN patients and induce constitutive activation of JAK2-STAT (signal transducer and activator of transcription) signaling, promoting proliferation, differentiation and survival of myeloid progenitors [2]. The most common mutation driving the development of MPN is JAK2-V617F [3–6]. In addition, a variety of mutations in JAK2 exon 12 (at amino acid positions 536–547) are pathogenic contributors to polycythemia vera (PV) [7]. Other JAK2 missense variants or insertions/deletions have been identified in MPN patients, but these are not well known, and their clinical significance remains unclear [1,8].

JAK2-I724T (rs372254348) in exon 17 has only a few reported instances in MPN; however, this may be due to the focus on identifying mutations in exons 12–15 because of their suspected link with pathogenicity [8]. We found three MPN patients with JAK2-I724T reported in the literature, one each with PV, essential thrombocythemia (ET) and primary myelofibrosis (PMF) [9–11]. Specifically, next-generation sequencing employing an 86-gene panel identified I724T in a 49-year-old female patient with PV out of 33 JAK2-V617F negative MPN patients [10]. This patient also had mutations in APC (adenomatous polyposis coli) and EZH2 (enhancer of zeste homolog 2) genes. In another study, next-generation sequencing with a 54-gene panel identified I724T alongside a CALR mutation in one PMF patient out of 101 PMF patients [9]. It is unclear whether JAK2-I724T was somatic or germline in these patients. In a third study, whole exome sequencing identified I724T as a germline variant in a 53-year-old female patient with ET out of 23 triple-negative ET patients [11].

The I724T variant is uncommon in general populations previously studied. The gnomAD_exome v2.1.1 data-set reports 22 JAK2 c.2171T > C (p.I724T) alleles, accounting for a frequency of 0.00008754. None of these alleles were homozygous. The gnomAD_genome v3.1.1 data set reports 20 JAK2 c.2171T > C (p.I724T) alleles, accounting for a frequency of 0.0001314, including one homozygous allele. East Asian ethnicity had the highest frequency in the gnomAD_exome v2.1.1 data set of 0.0005513 and European Finnish had the lowest frequency of 0.0. Unexpectedly, recent work identified an unusually high prevalence of approximately 14.9% of JAK2-I724T in healthy New Zealand Polynesian people (Māori and Pacific Islanders) (n = 56 out of 377 individuals tested) [12]. This presented a unique opportunity for us to examine if this germline variant associates with MPN. Previous work showed that New Zealand Polynesian patients are diagnosed with PV, ET and PMF at a younger age and die earlier with PV and PMF [13,14]. Thus, we hypothesized that JAK2-I724T may be contributing to these phenotypic differences and undertook to estimate the frequency of JAK2-I724T in Polynesian MPN patients compared with other New Zealand MPN patients.

Materials and methods

In silico analysis

The published literature (scopus.com) and databases of ClinVar (ncbi.nlm.nih.gov/clinvar/variation/134552), CIViC (civicdb.org), COSMIC v97 (cancer.sanger.ac.uk/cosmic), HGMD (hgmd.cf.ac.uk), and TCGA (portal.gdc.cancer.gov) were reviewed for reports on JAK2 c.2171T > C (p.I724T) rs372254348 and other JAK2 missense variants in MPN. For all variants found, gnomAD_exome allele frequencies (v2.1.1) were extracted and REVEL (Rare Exome Variant Ensemble Learner) scores were calculated using the dbNSFP v4.3 database (http://database.liulab.science/dbNSFP). REVEL is an Ensemble method for predicting the pathogenicity of missense variants. It integrates scores from approximately 40 other prediction algorithms, including SIFT and PolyPhen-2 [15,16]. REVEL scores >0.5 are more likely to indicate pathogenic variants and scores <0.5 are more likely to indicate benign variants [17].

To investigate the functional effects of the JAK2 variants on the JAK2 protein, I-Mutant 2.0 (folding.biofold.org/i-mutant/i-mutant2.0) software was employed using conditions of pH 7.0 and temperature of 37°C [18]. ChimeraX 1.4 was used to model and visualize the structural change of p.I724T and predict the impact on the JAK2 protein [19]. The JAK2 JH2 (Janus homology 2) pseudokinase domain (Protein Data Bank PDB code 4FVQ) [20] was aligned with both full-length dimerized JAK1 (PBD code 7T6F) [21] and TYK2 kinase/pseudo-kinase domains (PBD code 4OLI) [22].

Patient cohort

Patient studies were approved by the Central Health and Disability Ethics Committee (HDEC), approval numbers 16/CEN/92 and 16/CEN/91 granted on 8 August 2016 and 11 November 2016, respectively. The cohort included 69 patients from Auckland City Hospital (enrolled between 1991 and 2019) and 43 from Middlemore Hospital, which serves the largest Polynesian population in Auckland (enrolled between 2011 and 2019); however, these were not consecutive patients. Most patient samples were obtained from Te Ira Kāwai Tissue Biobank (n = 60), and other samples through Auckland City Hospital. Clinical and laboratory parameters were collected from medical records, including driver mutations (JAK2, CALR, and MPL), ethnicity, gender, and age at diagnosis. Driver mutations were tested centrally as a part of routine patient care by the Molecular Haematology Laboratory at LabPlus (Auckland City Hospital) according to the methods previously established [4,5,23–26]. Where molecular results were unavailable, additional tests were performed using equivalent methods (see below). Further clinico-pathologic details were obtained for patients identified to carry JAK2-I724T.

Detection of JAK2-p.V617F, CALR and MPL mutations

JAK2-V617F was tested using an allele-specific PCR as described before [4,5]. The following primers were used: forward-1 5’-AGCATTTGGTTTTAAATTATGGAGTATATT-3’ (the underlined nucleotide is mismatched to maximize allele discrimination), forward-2 5’-ATCTATAGTCATGCTGAAAGTAGGAGAAAG-3’ and reverse 5’-CTGAATAGTCCTACAGTGTTTTCAGTTTCA-3’. The forward-1 primer amplifies the mutant allele (203 bp product) and the forward-2 primer amplifies both the mutant and wild-type alleles (364 bp product). PCR amplification was performed in a 25 µl reaction volume containing 1x Platinum II PCR buffer, 1.5 mM MgCl_2, 0.2 mM dNTPs, 0.5 µM primers and 1.0 U Platinum II Taq Hot-Start polymerase (Invitrogen, Waltham, MA, USA). PCR cycling conditions included an initial denaturation for 2 min at 94°C, followed by 33 cycles of denaturation for 15 s at 94°C, annealing for 15 s at 58°C and extension for 15 s at 68°C.

For CALR mutation detection, primers used were forward 5’-ACAACTTCCTCATCACCAACG-3’ and reverse 5’-GGCCTCAGTCCAGCCCTG-3’ [23,24]. For MPL mutation detection, primers used were forward 5’-GGCTGGCTGGATGAGGG-3’ and reverse 5’-GTTTACAGGCCTTCGGCTC-3’ [25,26]. PCR amplification was performed as for JAK2-p.V617F but in a 50 µL reaction with less primers (0.3 µM) and 1.25x GC enhancer included. Cycling conditions were also similar to the JAK2-p.V617F method but annealing was at 60°C and 35 cycles were done. PCR reactions were purified using QiaQuick PCR purification kit (Qiagen, Hilden, Germany) and Sanger sequencing was carried out by Auckland Genomics.

Detection of the JAK2-p.I724T variant

Carriers of the germline JAK2 c.2171T > C (p.I724T) variant were determined by a TaqMan® SNP genotyping qPCR technique. The TaqMan® SNP genotyping assay consisted of pre-designed sequence specific forward and reverse primers and two minor groove binder probes fluorescently labelled with VIC™: 5’-ATACCATGGGTACCACCTGAATGCA[T]TGAAAATCCTAAAAATTTAAATTTG-3’ and FAM™: 5’-ATACCATGGGTACCACCTGAATGCA[C]TGAAAATCCTAAAAATTTAAATTTG-3’ to detect the T and C alleles (in brackets) (Applied Biosystems, Foster City, CA, USA). Reactions were run according to the manufacturer’s protocol with minor modifications. Briefly, each reaction contained 10–15 ng DNA and 2.5 μl of TaqMan® genotyping master mix. The reactions were run in duplicated 384-well plates using the following PCR conditions: pre-read stage at 60°C for 30 s, hold stage at 95°C for 10 min, PCR amplification for 40 cycles at 95°C for 15 s (denaturation), 60°C for 1 min (annealing, extension) and holding at 60°C for 30 s.

Allele fraction and co-occurrence of the JAK2- p.V617F and p.I724T variants

Carriers of the somatic JAK2- V617F and germline I724T variants were analyzed with Oxford Nanopore Technologies GridION sequencing of an amplicon that included both variants [27]. Long range PCR amplified 7049 bp products using the following primers: forward 5’-ACGTTGATGGCAGTTGCAGGTC −3’ and reverse 5’-TGGTGCAGGAAGCTGATGCCTATC −3, with the Expand High Fidelity PCR System (Roche Applied Science, San Diego, CA, USA). PCR products were amplified in 50 μl reactions with 100 ng genomic DNA extracted from bulk tumour samples. Cycling conditions included an initial denaturation of 94°C for 3 min, then 10 cycles of denaturation at 94°C for 15 s, annealing at 60°C for 30 s and extension at 68°C for 5 min. This was followed by 25 cycles using the same conditions but with 5 s added to the extension time per successive cycle. The final step included an extension at 68°C for 7 min. PCR products were visualized with SYBR Safe DNA gel stain (Thermo Fisher Scientific, Waltham, MA, USA) on a 0.8% agarose gel and purified using AMPure XP beads (Beckman Coulter, Indianapolis, IN, USA). Two ligation-based amplicon libraries were prepared from these samples using the LSK-114 NBD-96 kit (ONT, Oxford, UK) according to the manufacturer’s instructions. A MinION R10.4.1 flow cell (FLO-MIN114) was run on a GridION instrument (MinKNOW software v.22.07.9) for 72 h (all from ONT). Super high accuracy basecalling was done using Guppy v6.3.9 (ONT). Reads were filtered using Filtlong (GitHub repository, https://github.com/rrwick/Filtlong, 2017) to ensure both variant positions were on the same read and mapped to hg19 with Minimap2: fast pairwise alignment for long nucleotide sequences (https://arxiv.org/abs/1708.01492, 2017). SNV allele fraction was identified with the Medaka tool (https://github.com/nanoporetech/medaka, 2020) and variant co-occurrence was identified using a custom script. Alignments were viewed using the Integrative Genomics Viewer (IGV) [28].

Statistical analysis

Statistical analyses were performed using IBM SPSS Statistics software version 25. Data are presented as mean (standard deviation, SD) for parametric data, or median (interquartile range, IQR) for non-parametric data. Normality was assessed via Shapiro–Wilk test. Differences in categorical variables were compared using Pearson Chi-square test if n > 5 or Fisher’s exact test if n < 5. Mean differences of continuous variables were compared using independent samples Student t-test for parametric data, or Mann–Whitney U test for non-parametric data. A P value < 0.05 was considered statistically significant.

Results

In silico analysis of previously reported JAK2 missense variants in MPN

We did not find a compilation of JAK2 variants identified in MPN patients. Therefore, we conducted a careful analysis of published studies and publicly available cancer databases to document such missense variants together with their experimental validation (Table 1). In addition, Supplemental Table S1 lists JAK2 variants previously reported in MPN but not yet experimentally tested. We extracted population allele frequencies and calculated REVEL scores for each variant (results are included in Table 1 and Supplemental Table S1). Unexpectedly, we found that several JAK2 variants distributed along the gene had high REVEL scores (>0.5), suggesting pathogenicity. Previous work documented REVEL scores >0.5 for 75.4% of disease-causing mutations but only for 10.9% of neutral variants [17]. The JAK2 variants that were found to increase JAK2 kinase activity in cellular models include Y317H and H345L in the FERM domain; F556V, R564Q, L579F, V617I, C618R, L624P, V625F, E627A, R683G, and F694S in the pseudo-kinase domain; and R867Q, T875N, and R1063H in the kinase domain (Table 1). Such effects may impact the MPN phenotype. For example, patients with co-occurring JAK2-V617F and R1063H had higher white cell and neutrophil counts [29]. Other patients with JAK2-V617F and co-occurring JAK2 variants had a higher incidence of transformation to AML [30].

Table 1. JAK2 missense variants reported in patients with myeloproliferative neoplasms along with their functional validation. Variants are listed in the order of the position on the gene.

JAK2 variant	Patient diagnosis	Driver mutations and high-risk mutations	Patient numbers in study	Ref	Experimental validation/Risk association	REVEL score, I-mutant stability, gnomAD_exome allele frequency^1
Y317H FERM (exon 8)	PMF	CALR (E378fs*45)	1/443 MPN patients	[31]	Increase in proliferation, STAT-3 and STAT-5 and AKT phosphorylation in response to IL3 in 32D-hEpoR cells [31]	REVEL score: 0.293 I-mutant ΔΔG: −1.05 gnomAD_exome: 0.0000163486
H345L FERM (exon 8)	PMF (male)	CALR (type 1), IDH1	1/427 MPN patients	[32]	Increase in proliferation compared to wild-type in response to EPO in Ba/F3-EpoR cells [32] STAT5 was phosphorylated in the absence of EPO and had significantly higher phosphorylation in response to EPO in Ba/F3-EpoR cells [32]	REVEL score: 0.151 I-mutant ΔΔG: + 0.66 gnomAD_exome: 0.0
F556V Pseudo-kinase JH2 (exon 13)	ET (male)	None	1/57 Triple-negative (PMF and ET) patients	[33]	No increase in STAT5 transcriptional activity in γ2A cells [33] Increased STAT5 phosphorylation in Ba/F3-MPL cells [33] Modelling indicates disruption to T555/R715 bond in the nucleotide-binding loop and affects JAK2 secondary and tertiary structure [34] Increased: JAK2 activation phosphorylation (Y1007/8), STAT5 transcriptional activity, STAT5 phosphorylation in γ2A cells [34] Increased cytokine-independent proliferation in Ba/F3-EpoR cells [34]	REVEL score: 0.892 I-mutant ΔΔG: −3.07 gnomAD_exome: 0.0000123208
R564Q Pseudo-kinase JH2 (exon 13)	ET (male, paediatric)	None	1/4 family case study	[35]	Increased kinase activity, cell apoptosis inhibition, MPL and STAT phosphorylation in the absence of TPO, in Ba/F3-MPL cells [35] Increased JAK2 activation phosphorylation (Y1007/8) in patient platelets positive for R564Q [35] More sensitive to ruxolitinib than JAK2-V617F cases [35]	REVEL score: 0.692 I-mutant ΔΔG: −0.91 gnomAD_exome: 0.0000280525
	MPN	Unknown	2/20,000 MPN patients	[8]
R564L Pseudo-kinase JH2 (exon 13)	MPN	Unknown	3/20,000 MPN patients	[8]	In silico analysis suggests change to the V617 position and the JH1/JH2 interface [36] Transformed to secondary AML and significant for occurring at higher frequency than general population [30]	REVEL score: 0.668 I-mutant ΔΔG: −0.46 gnomAD_exome: 0.00006412
	PMF	None	1/1161 MPN patients	[30]
	PV	Unknown	1/93 PV patients	[3]
L579F Pseudo-kinase JH2 (exon 13)	MPN	Unknown	1/20,000 MPN patients	[8]	Increased: cytokine dependent proliferation, STAT3 and STAT5 phosphorylation, number of cells in S phase, in Ba/F3 cells [37] Modelling indicates the JH2 ATP binding affinity was changed [37] Increased JAK2 activation phosphorylation (Y1007/8) in γ2A cells (but less than V617F). When L579F and V617F co-mutants are expressed, Y1007/8 is decreased compared to V617F alone [38]	REVEL score: 0.172 I-mutant ΔΔG: + 0.09 gnomAD_exome: 0.0
	PV	Unknown	2/65 PV patients	[39]
S591L Pseudo-kinase JH2 (exon 13)	MPN	Unknown	1/20,000 MPN patients	[8]	In silico analysis suggests S591L will be a severely deleterious mutation and will cause constitutive activation of JAK2 [36] No increase in auto-phosphorylation on Y1007/8 when S591L was expressed in HEK293 T or γ2A cells and no increase in STAT5, ERK or AKT phosphorylation when expressed in Ba/F3-EpoR cells [40]	REVEL score: 0.742 I-mutant ΔΔG: + 0.81 gnomAD_exome: 0.0
V617I Pseudo-kinase JH2 (exon 14)	MPN	Unknown	1/20,000 MPN patients	[8]	Increased: cytokine-independent proliferation, STAT5 transcriptional activity, and STAT5 phosphorylation, in Ba/F3-EpoR cells [41] No increase in phosphorylation of STAT5, ERK or AKT after IL-3 deprivation, in Ba/F3-EpoR cells. Mild increase in phosphorylation of a STAT5 peptide [40]	REVEL score: 0.631 I-mutant ΔΔG: −0.56 gnomAD_exome: 0.0
	ET (female)	None	1/30 Triple-negative MPN patients	[42]
	ET (female)	MPL-W515S, TET2, SF3B1	1/1 ET case study	[43]
V617F Pseudo-kinase JH2 (exon 14)	MPN				Reference MPN driver mutation (rs77375493)	REVEL score: 0.881 I-mutant ΔΔG: −2.30 gnomAD_exome: 0.000351631
C618R Pseudo-kinase JH2 (exon 14)	PV	JAK2-V617F	1/60 MPN patients	[44]	In silico analysis suggests change to the V617 position and the JH1/JH2 interface [36] Increased STAT-5 transcriptional activity in γ2A cells [45] Increased: cytokine-independent proliferation, STAT5 and ERK phosphorylation and Y1007/8 auto-phosphorylation in the absence of cytokines in Ba/F3-EpoR cells [45]	REVEL score: 0.492 I-mutant ΔΔG: −1.48 gnomAD_exome: 0.0
	PV	JAK2-V617F	1/1 Case study	[46]
	MPN	Unknown	1/20,000 MPN patients	[8]
	PV	JAK2-V617F	1/39 MPN patients	[47]
	PV	JAK2-V617F	1/1 PV case study	[48]
	PV (male)	JAK2-V617F	1/8 PV case study	[49]
L624P Pseudo-kinase JH2 (exon 15)	MPN	Unknown	1/20,000 MPN patients	[8]	Increased: cytokine-independent proliferation, STAT3 and STAT5 phosphorylation, and number of cells in S-phase, in Ba/F3 cells [37] Modelling indicates the JH2 ATP binding affinity was changed [37]	REVEL score: 0.912 I-mutant ΔΔG: −1.73 gnomAD_exome: 0.0
V625F Pseudo-kinase JH2 (exon 15)	ET (female)	None	1/57 Triple-negative (PMF and ET) patients	[33]	Increased cytokine-independent STAT5 transcriptional activity and STAT5 phosphorylation; hypersensitive to IL-3 and TPO [33]	REVEL score: 0.894 I-mutant ΔΔG: −1.87 gnomAD_exome: 0.0
E627A Pseudo-kinase JH2 (exon 15)	ET	None	1/2035 MPN patients	[50]	Increased: kinase activity, STAT5 transcriptional activity, STAT5 phosphorylation and proliferation in the absence of EPO in γ2A and Ba/F3-EpoR cell lines [51]	REVEL score: 0.674 I-mutant ΔΔG: −1.22 gnomAD_exome: 0.0
R683G Pseudo-kinase JH2 (exon 16)	ET	None	1/2035 MPN patients	[50]	Increased: kinase activity, STAT5 transcriptional activity, STAT5 phosphorylation and proliferation in the absence of EPO in γ2A and Ba/F3-EpoR cell lines [51] Detected in 15% of patients with B-cell ALL [51]	REVEL score: 0.833 I-mutant ΔΔG: −0.41 gnomAD_exome: 0.0
F694S Pseudo-kinase JH2 (exon 16)	PMF	TET2	1/1161 MPN patients	[30]	Increased Y1007/8 auto-phosphorylation in HEK293 T and γ2A cells [40] Increased STAT5, ERK and AKT phosphorylation after IL-3 deprivation in Ba/F3-EpoR cells [40]	REVEL score: 0.702 I-mutant ΔΔG: −2.47 gnomAD_exome: 0.0
I724T Pseudo-kinase JH2 (exon 17)	ET (female, germline)	None	1/23 Triple-negative ET patients	[11]	No increase in STAT5 transcriptional activity in HEK293T cells [11]	REVEL score: 0.753 I-mutant ΔΔG: −3.06 gnomAD_exome: 0.0000875448
	PMF	CALR (deletion p.L367Tfs*46)	1/101 PMF and secondary MF patients	[9]
	PV (female)	APC, EZH2	1/33 JAK2-V617F negative MPN patients	[10]
R867Q Kinase JH1 (exon 20)	ET	None	1/2035 MPN patients	[50]	Increased: cytokine-independent phosphorylation of STAT-1, −3, −5, AKT and ERK and cell proliferation in Ba/F3-MPL cells, (but not in Ba/F3-EpoR cells) [52] Increased STAT-1, −3 and −5 transcriptional activity in γ2A cells [52]	REVEL score: 0.896 I-mutant ΔΔG: −1.23 gnomAD_exome: 0.00000546926
T875N Kinase JH1 (exon 20)	ET (female)	None	1/1 case study	[53]	Increased: Y1007/8 auto-phosphorylation, ERK phosphorylation and STAT-3 and STAT-5 transcriptional activity in HEK293T cells [53] Increased cytokine-independent cell proliferation in Ba/F3 cells [53]	REVEL score: 0.653 I-mutant ΔΔG: −0.94 gnomAD_exome: 0.0
R1063H Kinase JH1 (exon 24)	PV	Unknown	3/93 PV patients	[3]	Significantly higher white blood cell and neutrophil count in JAK2 R1063H and V617F double mutation patients [29] R1063H and V617F double mutation had increased STAT5 -dependent transcriptional activity compared to V617F alone in γ2A cells [29] R1063H and V617F double mutation had increased: Y1007/8 auto-phosphorylation, and STAT-5 and ERK phosphorylation compared to V617F alone in HEK293T cells [29] R1063H alone did not have increased STAT5 transcriptional activity or increased phosphorylation [29]	REVEL score: 0.276 I-mutant ΔΔG: −0.81 gnomAD_exome: 0.00436708
	MPN	Unknown	18/427 MPN patients (germline)	[32]
	MPN	JAK2-V617F	14/390 MPN patients	[29]
	ET (male)	None	1/34 Triple-negative MPN patients	[54]
	PV (male)	None	1/34 Triple-negative MPN patients	[54]
	PMF (female)	None	1/128 PMF patients	[55]
	PMF	JAK2-V617F	6/128 PMF patients	[55]
	ET	None	1/1161 MPN patients	[30]
	ET	JAK2-V617F	3/1161 MPN patients	[30]
	PV	JAK2-V617F	2/1161 MPN patients	[30]
	PMF	JAK2-V617F	2/1161 MPN patients	[30]
	PMF	JAK2-V617F, DNMT3A, ASXL1	1/1161 MPN patients	[30]
	ET	JAK2-V617F, RUNX1, IDH1	1/1161 MPN patients	[30]	Transformed to secondary AML [30]
	ET	IDH1	1/1161 MPN patients	[30]	Transformed to secondary AML [30]

¹ REVEL scores and gnomAD_exome allele frequency (v2.1.1) were extracted from the dbNSFP v4.3 database [15,16]; ΔΔG values were extracted using I-Mutant 2.0 software (folding.biofold.org/i-mutant/i-mutant2.0) [18].

JAK2-p.I724T in a cohort of 111 New Zealand MPN patients

The JAK2-I724T variant (REVEL score 0.753, Table 1) was recently found to be germline with an estimated frequency of 14.9% (56/377) in healthy Polynesian people of New Zealand [12]. To examine if this germline variant predisposes to MPN, we investigated its prevalence in a cohort of 111 New Zealand MPN patients. The cohort included 33 patients with PV, 51 with ET and 27 with PMF (Table 2). There were 27 Polynesian patients, including 12 Māori and 15 Pacific Islanders. The remaining 84 patients consisted mostly of New Zealand Europeans (n = 70). Most Māori patients had PV (7/12, 58.3%), and carried JAK2-V617F mutation (10/12, 83.3%) (Table 2 and Table 3); whereas ET was more common for Pacific Island and European patients. There were no Māori patients with CALR mutations but only 5 Māori patients had ET (n = 4) or PMF (n = 1). Out of 15 Pacific Island patients, 7 had ET, 3 had PMF, and CALR mutations were detected in 3 of these patients (20.0%) (Table 2 and Table 3).

Table 2. Patients with myeloproliferative neoplasms tested for JAK2-I724T.

	PV	P value^a	ET	P value^a	PMF	P value^a	Total	P value^b
Total n of patients for each disease	33		51		27		111
Ethnicity, n (%)^c
Māori	7 (58.3)	0.007^d	4 (33.3)	0.433^e	1 (8.3)	0.444^d	12 (10.8)	0.052^d
Pacific	5 (33.3)		7 (46.7)		3 (20.0)		15 (13.5)
European	21 (30.0)		31 (44.3)		18 (25.7)		70 (63.1)
Asian	0 (0)		8 (66.7)		4 (33.3)		12 (10.8)
Other	0 (0)		1 (50.0)		1 (50.0)		2 (1.8)
Median age (IQR) at diagnosis
Total patients	57 (23)		60.0 (24)		66.0 (19)		60.5 (22)
Polynesian	53.0 (26)	0.327^f	55.0 (30)	0.103^f	57.5 (20)	0.081^f	54.5 (23)	0.005^f
European	62.0 (21)		65.0 (22)		69.0 (12)		66.0 (19)
Gender, n (%)
Male	18 (54.5)	0.760^e	25 (49.0)	0.129^e	20 (74.1)	0.037^e	63 (56.8)	0.100^e,g
Female	15 (45.5)		26 (51.0)		7 (25.9)		48 (43.2)
Driver mutation and JAK2-I724T, n (%)
Total JAK2	32 (97.0)		32 (62.7)		17 (63.0)		81 (73.0)
JAK2 exon 12	1 (3.0)		–		–		–
JAK2-V617F	31 (93.9)	0.001^d	32 (62.7)	0.161^d	17 (63.0)	0.136^d	80 (72.1)	0.004^e
MPL	0 (0)		4 (7.8)		4 (14.8)		8 (7.2)	0.068^d
CALR	0 (0)		10 (19.6)		5 (18.5)		15 (13.5)	0.010^d
Triple negative	1 (3.0)		5 (9.8)		1 (3.7)		7 (6.3)	-
JAK2-I724T	1 (3.0)^h		1 (2.0)^h		0 (0)		2 (1.8)^h	-

^a analysed within the disease category.

^b analysed between disease categories.

^c percentages were calculated within each ethnic group, except for the ‘Total’ column where percentages were calculated of the total number of patients. P values were calculated with Asian and Other considered together.

^d Fisher’s exact test (2-sided).

^e Pearson χ² test.

^f Mann-Whitney U test.

^g gender was significantly different (P = 0.033) between ET and MF patients.

^h co-mutated with JAK2-V617F.

Table 3. MPN driver mutations associated with different New Zealand ethnic groups.

Ethnicity	Total	Driver mutation, n (%)
		JAK2-V617F	JAK2 exon 12	MPL	CALR	Triple-negative	JAK2-I724T
Māori	12	10 (83.3)	0	1 (8.3)	0 (0)	1 (8.3)	1 (8.3)^a
Pacific	15	11 (73.3)	0	0 (0)	3 (20.0)	1 (6.7)	1 (6.7)^a
Asian	12	9 (75.0)	0	0 (0)	1 (8.3)	2 (16.7)	0 (0)
European	70	49 (70.0)	1 (1.4)	7 (10.0)	10 (14.3)	3 (4.3)	0 (0)

^a co-mutated with JAK2-V617F.

All 111 patients were tested for the presence of germline JAK2-I724T using a TaqMan genotyping assay. The variant was detected in two out of 27 (7.4%) Polynesian patients, one Māori and one Pacific Islander. None of the MPN patients of other ethnicities (0/84) had the variant (Table 3). Both patients carrying JAK2-I724T were also positive for JAK2-V617F (Table 2 and Table 3). Of the JAK2-I724T positive patients one had PV, the other had ET. Both patients had some unusual clinical features. The PV patient was a 53-year-old Māori male who presented with a hemoglobin (Hb) level of 173 g/l without leukocytosis. He was managed with monthly venesections, but three years later was noted to have progressive anaemia (Hb 91 g/l), marked thrombocytosis (platelet count 1127 x10⁹/l), and splenomegaly. His bone marrow was markedly hypercellular with abnormal megakaryocytes, hyperplastic and mildly left-shifted granulopoiesis (blasts 3%), and suppressed erythropoiesis. Reticulin staining was graded at MF-1 with patches of increase. Together, these features met the criteria for post-PV myelofibrosis (MF). Venesections were discontinued, and the patient was managed with intermittent hydroxyurea. Five years later, he presented generally unwell with anemia (Hb 81 g/l), marked thrombocytopenia (platelets 18 x10⁹/l) and marked leukocytosis (white cell count (WCC) 49.7 x10⁹/l) with 23% blasts (11.43 x10⁹/l). Bone marrow showed 50–60% blasts that were myeloperoxidase positive, atypical megakaryocytes and increased reticulin staining at MF-3, indicating transformation to acute myeloid leukemia (AML) on the background of post-PV MF. The disease was refractory to chemotherapy and the patient died three months later. It was also noted that at the age of 57 years, this patient was diagnosed with small cell lung cancer, which was treated surgically.

The ET patient with JAK2-I724T was a 60-year-old Cook Island Māori female. She presented with frequent sweating and palpable splenomegaly. Her platelet count was 494 x10⁹/l, accompanied by mild leukocytosis (WCC 12 x10⁹/l) with mild neutrophilia (8 x10⁹/l). Hb was normal at presentation (135 g/l). Bone marrow was normocellular to mildly hypercellular with markedly increased numbers of hyperlobulated megakaryocytes, typical for ET. Reticulin staining was not increased. The patient was managed with intermittent hydroxyurea. Six years later, her platelet count was similar (473 x10⁹/l), but mild leukocytosis (WCC 13.7 x10⁹/l, neutrophils 10.4 x10⁹/l), and mild anemia developed (Hb 112 g/l). Overall, the presentation of this patient was typical for ET, although it was a high-risk disease with an International Prognostic Score for Essential Thrombocythemia (IPSET) of 3, which carries an expected median survival of 13.8 years [56].

To examine if JAK2-V617F was acquired on the same or different allele as the germline I724T variant, Oxford Nanopore sequencing was employed. The JAK2 region spanning p.V617 and p.I724 loci (7049 bp in length) was amplified from bulk tumour samples using long-distance PCR. The variant allele fraction (VAF) and the co-occurrence of p.V617F and p.I724T were determined in full-length reads. The PV patient had a VAF of 0.55 for p.V617F and 0.76 for p.I724T. The ET patient had a VAF of 0.33 for p.V617F and 0.48 for p.I724T. Both patients had p.V617F occurring on the same allele as p.I724T, as well as separately (Table 4). No other variants were detected in the sequence region spanning exons 14–17 of JAK2.

Table 4. Co-occurrence of JAK2 V617F and I724T.

	Patient 1 – PV		Patient 2 – ET
	Number of reads	VAF	Number of reads	VAF
V617F and I724T	3064	0.45	599	0.12
V617F only (I724 wild-type)	641	0.09	1041	0.21
I724T only (V617 wild-type)	2110	0.31	1792	0.36
Both wild-type	988	0.15	1546	0.31
Total reads	6803	1	4978	1
Other calculated combinations
I724T with or without V617F	5269	0.76	2439	0.48
V617F with or without I724T	3748	0.55	1645	0.33
V617 wild-type with or without I724T	3127	0.45	3373	0.67
I724 wild-type with or without V617F	1629	0.24	2587	0.52

VAF, variant allele fraction.

The precise genotype of cancer cells cannot be determined in bulk tissue. However, the mixture of reads carrying p.V617F and p.I724T together and separately implied the presence of cancer subclones in both patients. In the PV patient, the dominant clone appeared to be homozygous for both p.V617F and p.I724T. The p.I724T VAF of 0.76 (instead of around 0.5) indicated that the loss-of-heterozygosity (acquired uniparental disomy) had occurred in a proportion of the cells, which is common on chromosome 9p in PV patients [57]. In the ET patient, dominant cancer clones appeared heterozygous for p.V617F, but some also carried p.I724T (VAF 0.12) (Table 4).

In silico analysis of the impact of I724T on the JAK2 protein

All JAKs respond to cytokines through interaction with cytokine receptors such as thrombopoietin (TPO)-receptor (MPL). JAK2 has four major domains: a FERM (four-point-one, ezrin, radixin, moesin) domain, an SH2 (Src homology 2) domain, a JH1 (Jak homology 1) tyrosine kinase domain and a JH2 pseudo-kinase domain (Figure 1A). The JH2 pseudo-kinase domain is involved in autoinhibition of the JH1 kinase domain in the absence of ligand–receptor interaction. The FERM and SH2 domains associate with the intracellular domains of cytokine receptors, and upon cytokine binding, two JAKs dimerize to create an active signaling complex [58]. Dimerization enables transphosphorylation of Y1007 and Y1008 in the activation loop of the JH1 kinase domain, increasing catalytic activity. Activated JAK2 phosphorylates tyrosine residues on the cytokine receptor intracellular domains, creating a docking site for STAT molecules. STATs bound to the cytokine receptor-JAK2 complex are subsequently phosphorylated by JAK2, which leads to STAT dimerization and translocation to the nucleus, where they initiate transcription of cytokine-responsive genes, promoting cell proliferation and survival. In addition, JAK2 also activates the PI3K/PKB/mTOR pathway as well as the MAPK pathway, including RAF, MEK1/2, and ERK1/2 [59].

Figure 1. Modeling of JAK2-I724T. (A) JAK2 schematic depicting its four major domains: a four-point-one, ezrin, radixin, moesin (FERM) domain, a Src homology 2 like (SH2) domain, a pseudo-kinase Jak homology 2 (JH2) domain, and a tyrosine kinase (JH1) domain. The start and end amino acid positions are indicated for each of the domains, as are the mutant/variant positions (617 and 724). (B) Human JAK2 JH2 domain (PBD code 4FVQ) was superimposed onto the JAK1 JH2 domain of activated, dimerized mouse JAK1 (PBD code 7T6F). The structure was visualized at full-length (Bi), as well as zoomed into the dimerized JH2-domains alone (Bii). One of the JAK1 monomers is colored pink and the other is colored purple. The JAK2 JH2 domain is colored blue. The amino acid position of I724 of JAK2 and the corresponding amino acid V763 in JAK1 is colored red. JAK2 V617 is colored in orange, and the corresponding amino acid 657 in JAK1 is colored in green. (C) Human JAK2 JH2 domain (PBD code 4FVQ) was superimposed onto the autoinhibited TYK2 JH2-JH1 structure (PBD code 4OLI). The TYK2 JH2 domain was hidden from view to enable easier visualization of the amino acid residue interactions. The TYK2 JH1 domain is colored pink and the JAK2 JH2 domain is colored light blue. JAK2 V617 is colored in orange (Ci). The bonds that were within 5 angstroms of I724 are colored in yellow. I724 (colored red), on alpha helix EF, contacts amino acids 721–726 (PECIEN) as well as I716 (on linker adjacent to the activation loop alpha helix), and L763 and Y766 (both on alpha helix G) (Cii). I724 was altered to a threonine and T724 makes contact with amino acids 721–726 (PEC(T/I)EN) and L763, but does not contact I716 and Y766 (Ciii).

To identify potential changes to the JAK2 protein harbouring an I724T substitution, an in-silico analysis was performed. I-Mutant 2.0 software was used to predict the change in the thermodynamic stability of JAK2-I724T and other JAK2 variants previously reported in MPN patients (Table 1, Supplemental Table S1) [18]. The difference in Gibbs free energy of folding (ΔΔG) between the wild-type I724 and variant T724 proteins identified that JAK2-I724T had a score of −3.06 kcal/mol, suggesting that I724T reduces the stability of the JAK2 protein. The majority of other JAK2 variants listed in Table 1 also had reduced stability, including JAK2-V617F (−2.30 kcal/mol) with JAK2-I724T and JAK2-F556V having the lowest scores (−3.06 and −3.07 kcal/mol, respectively). To further understand the potential effect of I724T on JAK2, ChimeraX software was used to visualize the variant impact (Figure 1). Both the V617F and I724T variants constitute part of the JH2 pseudo-kinase domain (Figure 1A). I724 is part of an alpha helix (Figure 1B, C), but the I724T variant is unlikely to disrupt this secondary structure due to the size similarities between the two amino acids and the hydrogen bonds being retained. However, the stability of the tertiary structure may be altered due to the change from a hydrophobic isoleucine to a hydrophilic threonine.

Full-length JAKs have been recalcitrant to structural analysis and only the structure of the JH2 pseudo-kinase domain is available for JAK2 [21]. Thus, to observe any interaction changes of I724T with other JAK domains, the JAK2 pseudo-kinase domain (PBD code 4FVQ) was aligned with both dimerized activated JAK1 (PBD code 7T6F) (Figure 1B) and autoinhibited TYK2 kinase/pseudo-kinase domains (PBD code 4OLI) (Figure 1C). When JAK1 was dimerized, V617 from each monomer closely interacted with each other, whereas the two I724 amino acids were further away from each other (Figure 1Bi, Bii). In the autoinhibited model, the I724 residue was not in contact with the JH1 kinase domain (Figure Ci). When amino acid position I724 was altered to a threonine, T724 no longer had contact with I716 and Y766 (Figure 1Cii, Ciii). I716 is adjacent to R715 at the end of the activation loop, which is conserved in the JAK family. T555 is part of the nucleotide-binding loop and makes a hydrogen bond with R715 both in the apo and ATP-bound form of JH2 [20]. However, the I724T variant did not appear to affect the interaction of R715 and T555. Within the JH2 domain, autophosphorylation of Y637 has been shown to increase JAK2 activity, and autophosphorylation of Y570 decreases JAK2 activity [60,61]. In conclusion, our in-silico analysis suggested that I724T was unlikely to provide JAK2 a gain-of-function in which to drive MPN pathogenesis.

Discussion

This study investigated the presence of a germline JAK2-I724T variant in a cohort of 111 New Zealand MPN patients, including 27 Polynesian patients. We found that the prevalence of germline JAK2-I724T (2/27; 7.4%) in Polynesian patients was higher than that reported in the gnomAD_exome v2.1.1 dataset (0.00008754) or in any other cohort of MPN patients previously published, but lower than expected compared to healthy Polynesian people (56/377; 14.9%). The two JAK2-I724T positive MPN patients were also positive for V617F, arguing against I724T being a primary MPN driver. Congruently, in silico analysis demonstrated that I724T is unlikely to give JAK2 a gain-of-function.

Both patients carrying JAK2-I724T in our cohort had poor-risk features. The PV patient was young (53 years at the time of diagnosis and 61 years at death), compared with the median age of PV patients of approximately 65 years at diagnosis and 81 years at death [62,63]. This patient progressed to bone marrow fibrosis at 3 years after diagnosis, and to AML at 8 years, which is expected for less than 10% and less than 5% of patients, respectively [64]. Usually, adult PV patients who present at 61 years of age or younger have a median survival time of 23 years [65]. Progression to MF occurs in 12–21% of PV patients and approximately 7% develop AML within 20 years [66]. Similarly, our ET patient with I724T had unfavourable signs of leukocytosis, splenomegaly and anemia. Leukocytosis confers a higher risk of thrombosis [67], and anemia is an independent risk factor for inferior survival in ET [68]. Splenomegaly is not currently included in the major prognostic models that estimate patient survival in ET, but its presence was found to be associated with an increased risk of thrombosis and poorer survival [69–71]. Previous work documented that New Zealand Polynesian patients with MPN have several distinctive features, including lower age at diagnosis and higher mortality rates [13,14]. Multiple factors likely influence these disparities and these remain largely unknown. Our study was small, with only two of the 111 MPN patients studied harboring the germline JAK2-I724T variant. Further work is required to establish the true incidence of this variant in a larger cohort of Polynesian patients and exclude its role in disease. It remains possible the JAK2-I724T variant may modulate JAK2 activity and consequently, the MPN phenotype.

Using Nanopore sequencing, we demonstrated that both of our JAK2-I724T-positive patients carried V617F on the same allele, as well as separately, indicating subclonal disease. A recent study elegantly dissected the clonal architecture in MPN by sequencing DNA from single-cell-derived colonies. The presence of subclones and co-occurrence of JAK2-V617F with high-risk mutations was predictive of poor patient outcomes [72]. We have not tested other mutations in this study, but the presence of JAK2- V617F and I724T subclones may align with the high-risk MPN phenotype in our patients. Our results emphasize the need to examine the potential co-operativity between JAK2- V617F, I724T and other MPN mutations in Polynesian patients, in particular those with high-risk features. Because of the small sample size, no inference can be made on the relationship between the VAFs of these variants and the MPN severity.

The in silico analysis we conducted highlights that non-canonical JAK2 variants with REVEL scores >0.5 are more common in MPN than clinically realized. Some of these variants impact JAK2 activity, cell proliferation in vitro, and the MPN phenotype in patients, implying potential clinical relevance. JAK2-V617F has a REVEL score of 0.881, and JAK2-I724T has a REVEL score of 0.753. Typically, a higher REVEL score (closer to 1) suggests a higher probability that the variant is pathogenic or disease-causing. Conversely, a lower REVEL score (closer to 0) indicates a lower probability of pathogenicity. However, using REVEL score has limitations, and there is not a universally agreed-upon threshold for determining variant pathogenicity. Previous work showed that REVEL scores >0.5 identify 75.4% of disease-causing mutations but are also associated with 10.9% of neutral variants [17]. We found that several JAK2 variants functionally confirmed to be pathogenic had low REVEL scores, such as Y317H (0.293), H345L (0.151), L579F (0.172) and R1063H (0.276) (Table 1). This emphasizes that while useful, REVEL scores should be interpreted with caution. Experimental validation and consideration of clinical data will be crucial for determining the clinical significance of various JAK2 variants.

Similar limitations apply to the use of the ΔΔG score. I-Mutant 2 is a bioinformatics tool that predicts the impact of amino acid substitutions on protein stability. The output it provides is in the form of a ΔΔG value. A negative ΔΔG value indicates that the mutation is predicted to be destabilizing for the protein, meaning it is likely to decrease the stability of the protein structure. The larger the absolute value of the ΔΔG, the more significant the predicted impact on protein stability [18,73]. We found that JAK2-V617F had a score of −2.30 kcal/mol and JAK2-I724T had a score of −3.06 kcal/mol, suggesting that both variants have a moderately destabilizing effect on the JAK2 protein. However, this prediction is based on computational models, and experimental validation is needed to demonstrate the actual impact of I724T on the stability of the JAK2 protein.

Although the REVEL score of 0.753 and the ΔΔG value of −3.06 kcal/mol suggested that JAK2-I724T could be pathogenic, ChimeraX analysis did not identify obvious structural changes that would give the protein a gain-of-function. In accordance, when Inano et al overexpressed JAK2-I724T together with a STAT5 response element in HEK293T cells, there was no increase in the level of STAT5 activation [11]. Although HEK293T cells are not hematopoietic, they have been used to study STAT5 activation [53]. To clarify the impact of I724T on JAK2 activity, it would be of interest to identify whether JAK2-I724T increases STAT5 phosphorylation, or if it has increased activation via autophosphorylation using hematopoietic Ba/F3 or γ2A cell lines [33,37,45,52]. It would also be of interest to determine whether JAK2- I724T and V617F co-variants have increased JAK2 activity compared to JAK2-V617F alone. HEK293T cells expressing co-occurring JAK2- V617F and R1063H had increased JAK2 activity compared to JAK2-V617F alone, yet JAK2-R1063H alone had no increase in activity, similar to JAK2-wild-type [29]. In addition, patients with co-occurring JAK2- V617F and R1063H (n = 14) had a higher risk MPN phenotype [29]. Future work could include STAT5 mRNA expression comparisons between patients carrying I724T and those who don’t. STAT5 transcripts are highly expressed in PV patients compared to ET and PMF, regardless of JAK2-V617F status, but are especially high in JAK2-V617F-positive patients [74]. Comparing STAT5 transcript expression levels between patients with and without I724T, as well as between ethnic groups, and comparing this to disease severity may help identify potential genetic links influencing disease activity in patients.

The true incidence of I724T and of other JAK2 variants (germline or somatic) in MPN is unknown, and further work is needed to determine their roles in disease. It has been proposed that inherited mutations or congenital predisposition provide the requisite genetic perturbation for the development of MPN, justifying further studies into the role of germline JAK2 variants in MPN [63,75,76]. For most JAK2 variants found in the literature the germline status is unknown. In contrast to our study, the three I724T-positive MPN patients reported previously were negative for JAK2-V617F [9–11]. One patient carried CALR mutation [9] and two were triple-negative [10,11]. Only a few studies have investigated the prevalence of non-canonical JAK2 variants by sequencing the whole of JAK2. Most studies have only focussed on exons 12–15, thus the variants reported are clustered to this region. The more recent studies have focussed on the whole of JAK2 e.g. [9–11,30,50,55,77]. Future studies are needed to clarify the types and frequencies of JAK2 variants in MPN. It is important to identify which of these variants contribute to disease pathogenesis, and which do not. Two or more pathogeneic mutations in JAK2 may confer additive proliferative advantage or indicate genomic instability with higher mutagenic probability, which impacts disease progression and can act as a marker of poorer prognosis [30,78].

New Zealand Polynesian patients with MPN fare worse than their European counterparts [13,14]. While health system issues may contribute to worse outcomes, lower age of diagnosis suggests biological drivers may play a role, so it is possible that mutational load differences influence distinctive MPN phenotype in Polynesian patients. There are occasional case reports describing the presence of JAK2-I724T in other hematologic malignancies including AML [79] and acute lymphoblastic leukemia [80]. It may be of interest to determine the prevalence of I724T in Polynesian patients with such diseases as well.

Conclusion

The prevalence of germline JAK2-p.I724T in our cohort of 111 MPN patients is the highest reported in the literature thus far; however, we found no evidence that this variant associates with MPN. We conclude that JAK2-I724T does not drive MPN pathogenesis, but we cannot exclude its contribution to the MPN phenotype. Due to the high prevalence of JAK2-I724T in the general New Zealand Polynesian population and the well-documented high impact of MPN in this ethnic group, studies to clarify these possibilities should be prioritized. From a global perspective, Polynesian patients with MPN provide a unique cohort in which to study the potential role of JAK2-I724T in MPN development and progression. Our in silico analysis of other JAK2 variants in MPN highlights the need to further understand the relationship between JAK2 mutational load and the disease process. Sequencing of the whole of JAK2 in MPN patients, in particular those with high-risk disease, would help determine the role of JAK2 variants in MPN pathogenesis.

Author contributions

CP and TNG tested patient samples and analyzed results; TI conducted bioinformatic analyses and drafted the manuscript; PT supported analysis of the Nanopore data; PRS provided supervision and advice to CP; MLK-Z designed the study, helped with data analysis and interpretation, and wrote the paper. All authors have read and agreed to the version of the manuscript that has been submitted.

Acknowledgements

The authors thank Alice Rykers, Anona Pak and Aparajita Chatterjee (Te Ira Kāwai Tissue Biobank) for providing patient samples with the pertinent clinical information; Kristine Boxen and Paula Shields (Auckland Genomics) for performing Sanger sequencing and Oxford Nanopore sequencing respectively; Nikhil Ghallyan (LabPlus, Auckland City Hospital) for sharing methodology references; and Marie-Christine Morel-Kopp (Royal North Shore Hospital, Sydney) for her advice on bioinformatic analysis and reviewing the paper prior to submission.

Disclosure statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Data availability

All data generated and analyzed during this study are included in this article.

Additional information

Funding

CP was funded by an Health Research Council of New Zealand Pacific PhD Scholarship. The work was supported by funding from Auckland Medical Research Foundation (funder reference 1119009), Bone Marrow Cancer Research Trust (Christchurch; UoA grant number 3720536) and Cancer Research Trust (UoA grant number 3720177) to MLK-Z.