Reconciling Phenotype with Genotype in the MPN: Impact of SNP Array-Based Chromosomal Analysis.

Abstract 1893

Poster Board I-916

The diagnosis and accurate prognostication of the myeloproliferative neoplasms (MPN) is complicated by phenotypic mimicry and variable rates of disease evolution. The somatic mutation of JAK2 (JAK2V617F), identified in more than 90% of PV and in 50% of ET and PMF patients, is associated with acquired uniparental disomy (aUPD) on chromosome 9p, generating copy number-neutral loss of heterozygosity (CN-LOH) and results in homozygosity of JAK2V617F in roughly one third of JAK2V617F-positive patients. The molecular events upstream of the mitotic recombination that leads to somatic CN-LOH are essentially unknown. Intriguingly, JAK2V617F itself has been implicated in the generation of genomic instability (Plo et al 2008). Although these observations provide a rationale for how a single mutation could give rise to different clinical pathologies and downstream genomic instability in MPN, the JAK2V617F allele burden (AB) has not been systematically correlated with high resolution evaluation of amplifications, deletions, and CN-LOH in a large MPN cohort. We performed single nucleotide polymorphism arrays (SNP-A), a powerful karyotyping tool with the unique ability to detect CN-LOH on neutrophil DNA from 90 MPN patients. Published copy number variants (CNVs) and those identified in our internal cohort of 995 healthy controls were excluded as well as germline, non-clonal CN-LOH regions based on size criteria. Genomic results were correlated with the quantitative JAK2V617F AB, clinical phenotypes within and between disease classes, and in 21 paired longitudinal samples. SNP-A detected aUPD and/or chromosomal gains or losses in 81% of MPN patients. The prevalence of genomic lesions was lowest in ET compared to PV or PMF (40% of ET cases were lesion free), and lower still in JAK2V617F-negative ET (62% lesion free). aUPD was the most common genomic lesion, occurring in 59% of the MPN cases, and involving every chromosome except chromosomes 18 and 23. As expected, aUPD most commonly involved 9p (38 cases), encompassing JAK2. aUPD encompassing the TET2 gene at 4q24, recently implicated in MPN and other myeloid disorders, was seen in 3 of the 90 cases, with deletions spanning this gene in 2 additional cases. Five cases of aUPD at 2q were observed, and other recurrent regions were limited to 3 cases or fewer. The prevalence of chromosomal losses alone was far lower than UPD (22%), most commonly involved chromosome regions 20q, 13q and 17q, and was a feature primarily of PMF. The prevalence of aUPD in our MPN series of 59% was higher than in MDS (20%, p<0.001), MDS/MPN (35%, p=0.0074), and sAML (23%, p=0.0004) using the same arrays and analytical methods. Independent of MPN classification category, JAK2V617F positive patients had a higher prevalence of UPD than did JAK2V617F negative patients, although when excluding 9p UPD (24/59 vs 11/31) this was not significant. However, homozygous JAK2V617F patients had twice the prevalence of aUPD (again excluding 9P UPD) than did the heterozygous patients (8/25 vs 5/33), and twice the prevalence of multiple aUPD within individual cases. Of the 21 patients studied longitudinally over 5 years (range 2–9), 3 patients with chromosomal lesions present on the first sample were not present in the second: 9p UPD resolved in one PV patient who transformed to PMF, 1p UPD resolved in a JAK2V617F-negative ET patient, and a 20q deletion resolved in a heterozygous JAK2V617F-positive ET patient who transitioned to PV. Additional UPD regions occurred in only 1/8 heterozygous JAK2V617F positive MPN patients, while 5/11 JAK2V617F homozygous patients developed additional aUPD. In contrast, the acquisition of deletions occurred primarily in JAK2V617F heterozygous ET patients, all of whom had transitioned to PV or PMF. We conclude that aUPD is increased in the MPN compared to other myeloid disorders, and associates with the presence of JAK2 V617F in a gene dose-dependent manner. We hypothesize that aberrant JAK2 signaling contributes to the development of UPD but not chromosomal losses or gains. Furthermore, comparison of deletion, amplification, and UPD events in ET and PV relative to PMF suggest that PMF bears a closer genomic resemblance to MDS than to ET or PV. Thus SNP-A provides unique insight into MPN disease classification (PMF), disease progression, and genomic instability mechanisms.