Large Genomic Alterations Occurring in the Transition from Chronic to Blast Phase of Chronic Myeloproliferative Neoplasms

INTRODUCTION Five to 20% of patients with myeloproliferative neoplasms (MPN), including Essential Thrombocythemia (ET), Polycythemia Vera (PV) and Primary Myelofibrosis (PMF), transform to an aggressive secondary acute myeloid leukemia (sAML). While several studies reported the association of some mutations with the risk of leukemic transformation (Vannucchi AM et al, Leukemia 2013), the mechanisms that contribute to transformation from MPN to sAML remain largely poor characterized.

METHODS. We collected annotated samples from 15 chronic-phase (CP) MPN patients (pts), 6 pts with accelerated phase (AP; PB blasts 10-19%) and 12 pts with sAML; for the latter, paired samples (chronic/blast phase-BP) were available. CP and BP samples were separated by a mean of 77 (12 to 216) months interval. We used Illumina whole exome sequencing (WES) to identify copy number variations (CNV) in all samples. In the paired samples set, we also performed long reads genome sequencing by the Oxford Nanopore technology, a uniform process that generates sequences randomly and independently, without classical sources of bias such as GC-content and mappability. Data analysis for CNV detection was performed by a novel devised computational package (Nano-GLADIATOR; Magi A, Bartalucci N et al, Genome Biology, submitted) allowing the analysis of individual samples without the need of paired-analysis.

RESULTS. The mean number of CNV detected in CP, AP and BP samples was respectively 130.7±49, 132±42 and 177.4±61 (P=0.03 of BP vs CP). CNV were represented by gain of genomic material in 63.6%, 68.2% and 64.1% of CP, AP and BP. Considering the length of all CNV, expressed as base pairs (bp), we found that 96.8% of CNV in CP were focal alterations spanning <1 Mega bases (Mb) while only 2.9% and 0.3% were larger than 1Mb and 30Mb, respectively. Conversely, alterations involving >1Mb and >30Mb in BP samples were 14.4% (P=0.05) and 2.1% (P=0.04), and corresponding figures in AP were respectively 5.3% (ns vs CP, P=0.05 vs BP) and 0.7% (P=0.04 vs CP). Considering CNV >1Mb only, 53.7% were gains in CP compared with 68.1% in AP and 63.6% in BP, while losses were 46.3%, 31.9% and 36.4% respectively. Furthermore, alterations involving all the short (p) and long (q) -arm or the whole chromosome were found in 77% of BP compared to 36% of AP and 11% only of CP samples (P<0.01), overall indicating that larger alterations are enriched in BP. We used Nanopore platform to analyze paired CP and BP samples. We found that the total number of bp involved in CNV was 380x10⁶ and 2x10⁹ in CP and BP, respectively, with an estimated frequency of 0.01 and 0.07 altered base every 1 base of normal genome. The mean length of CNV was 5.6x10⁶ bp and 23.7x10⁶ bp respectively in CP and BP samples. There was a total of 29 new CNV acquired in BP samples compared to CP, involving 10 different chromosomes. We identified recurrent alterations as double or single deletion of 100,000 bp in chromosome 8 (55.5% of BP), deletion of a 510.000 to 610.000 bp region of chromosome 4q (40% of BP) and an amplification of 32,535,000 bp in chromosome 21q in 34% of BP samples. Alterations involving chromosomes 4 and 8 detected in BP samples were already present at paired CP samples in 75% and 80% of cases, while in the remaining 25% and 20% they were acquired at BP. The same abnormalities were absent from the 15 unpaired CP samples (WES data) whereas they were present in 4/6 and 3/6 of AP patients, respectively.

CONCLUSION All together, this data indicate that genomic instability is a hallmark of leukemic transformation of MPN and for the first time identify regions that may be recurrently associated with disease progression from CP to BP, potentially representing novel biomarkers. These finding require confirmation in larger series.