Single Nucleotide Polymorphism Array (SNP-A) Karyotype Is the Best Predictor of Prognosis In Normal Cytogenetics Acute Myeloid Leukaemia (AML)

Abstract 411

The herald of next generation sequencing has ushered an era for the discovery of frequent novel mutations such as DNMT3a, IDH1/2 and ASXL1 in acute myeloid leukaemia with a normal karyotype (AML-NK). The frequency of these mutations has been extensively documented; however, there is inconclusive evidence for their prognostic significance. Furthermore, whilst having a ‘normal' cytogenetic karyotype, AML-NK has been characterized by SNP-A karyotyping revealing multiple cryptic genomic aberrations. To predict prognosis, mutations of FLT3, NPM1 and CEBPA as well as gene expression profiles have been used as biomarkers. Using 454 next generation sequencing (NGS), we sought to determine the prognostic significance of mutations in ASXL1, DNMT3a, FLT3, IDH1/2, NPM1 and TP53 as well as detect cryptic genomic aberrations using 250K SNP microarrays in 92 AML-NK patients (pts) uniformly treated from the UK MRC/NCRI AML clinical trials, excluding pts with induction deaths.

We demonstrate the presence of 54 cryptic genomic aberrations in 37 pts consisting of 11 deletions (del) (8 pts), 10 gains (10 pts) and 33 regions of UPD (29 pts). Aberrations were excluded as CNV's if there was >50% overlap with variants from the DGV and an internal cohort of 91 healthy subjects. The median size of del was 1.6Mb(0.18Mb-3.4Mb) with chromosome (chr) 4(q24) and 17(q11.2 and q22.1) deleted in two pts. One pt had micro del on chr12p13.2-p12.3 and 21q22.11-q22.12 affecting the ETV6 and AML1 genes, respectively, that was confirmed by FISH. Gains (median 0.37Mb(0.26–0.71Mb)) were identified on chr8 (4 pts). The median size of UPD's was 83.3Mb(20.5Mb-158Mb) with interstitial UPD <20Mb excluded from the analysis. Chromosome 13 (q12.11-q34) and 6(p25.3-p22.1) were the most frequently affected with UPD observed in 15 and 5 pts, respectively. Of the 15 pts with UPD of chr13, 11 pts had this occurrence in small clones. Additional regions affected by UPD included chromosomes 1, 2, 4, 5, 7 and 11.

NPM1, DNMT3a, FLT3-ITD, IDH2, FLT3-TKD, IDH1, TP53 and ASXL1 mutations occurred in 64%, 49%, 22%, 5%, 3%, 2% and 1% of patients, respectively with 13 cases having no detectable mutations. The R882 mutation hotspot of DNMT3a occurred in 33/45 cases with clone size ranging from 34–65% and occurred concurrently with mutations in NPM1 (31 pts), FLT3-ITD/TKD (30 pts) and IDH2 (9 pts). IDH1 (R130H) and IDH2 (R140Q and R172K) mutations were mutually exclusive and occurred in 23 pts, with 4 pts having concurrent FLT3, NPM1 and DNMT3a mutations. Only 1 patient had an ASXL1 mutation with a concurrent IDH2 mutation and 2 cases had mutation of TP53. FLT3-TKD/ITD mutations were exclusive of each other. Not all pts with FLT3-ITD mutation had UPD of chromosome 13. There was no correlation between genomic aberrations and mutations of any other assayed genes.

The median duration of follow-up from time of diagnosis of the study group was 44 months (range 2–160). There were no significant correlations between the presence of genomic aberrations and age, sex, WBC, secondary disease or mutations in IDH1/2, FLT3-ITD or DNMT3a. Patients with FLT3-TKD were likely to have gains (2/5) compared to those with wild type (8/85, p=0.04). Patients with genomic aberrations had worse overall survival (OS) compared to pts without aberrations, (5 year OS 12% vs 40%, hazard ratio (HR) adjusted for other prognostic variables 2.56 (1.43–4.60), p=0.001); this was also seen when considering UPD on OS ((8% for those with UPD compared to 37% for those without UPD (adjusted HR 2.52(1.40–4.54) p=0.002) and cumulative incidence of relapse (CIR) (78% in pts with UPD compared to 54% for those without UPD (adjusted HR 4.04(1.84–8.90) p=0.0003). Adjusted analysis showed a dose effect for the number of UPD's on OS (p=0.007) and relapse (p=0.001). In a model building analysis, adjusted for age, WBC, sex, secondary disease, performance status and mutations of IDH1, IDH2, FLT3-ITD/TKD, DNMT3a and NPM1, only UPD proved independently prognostic for relapse and survival. We conclude that cryptic genomic aberrations in uniformly treated AML-NK pts provide better prognostic value and outcome prediction than metaphase cytogenetics and mutation detection, thereby, helping to identify patient groups with poor prognosis and requiring specific therapeutic strategies.