Chromosomal defects Detected by SNP-Array-Based Karyotyping Are Independent Predictors of Survival in Acute Myeloid Leukemia (AML)

Cytogenetics is the most important predictor of outcomes in AML. Traditional metaphase cytogenetics (MC), can detect abnormalities in only 40–60% of AML patients. Whole genome scanning by single nucleotide polymorphism arrays (SNP-A) can identify somatic chromosomal changes in hematopoietic malignancies and, due to its superb resolution, may detect previously cryptic unbalanced defects, even in samples deemed “normal” or uninformative using MC. Through simultaneous detection of loss of heterozygosity (LOH) and gene copy number changes, SNP-A also facilitate the identification of somatic segmental uniparental disomy (UPD). Here we tested whether SNP-A analysis could improve the detection rate of chromosomal defects in AML and enhances the prognostic value of MC. Analyses were performed using 250K and/or 6.0 Affymetrix SNP arrays on 140 primary (p) and secondary AML (sAML) patients (newly diagnosed= 107, relapsed=15, remission= 12, persistent=6) and 116 healthy controls. Data on cytogenetic detection rate, complete remission (CR), overall survival [OS], relapse free survival [RFS], remission duration [RD], and event free survival [EFS]) rates were obtained from patients who received induction chemotherapy. We also performed Flt-3 ITD, Flt-3 TKD and NPM-1 mutation analysis and integrated the clinical outcomes with SNP-A results. For patients in whom new defects were detected, germ-line DNA was also analyzed whenever technically possible. The cytogenetic abnormality detection rate in patients with active disease was higher with SNP-A compared to MC (pAML, 75% vs 43% p=<0.0001; sAML, 81% vs 53% p=0.0015). UPD comprised a significant proportion of the SNP-A detected defects (36% in pAML and 40% in sAML) and included chromosomal defects not described in a previous 10K SNP study, such as 1p, 3p, 7q, 11q, 13q, 17q, 20, and 21q. Newly diagnosed AML patients with SNP-A lesions had less favorable outcomes. This was true for all AML patients (OS [5.8 months vs not reached {NR}, p=<0.0001], RFS [6.4 months vs NR, p=0.04] RD [6.9 months vs NR, p=0.04], EFS [2.7 vs 17 months, p=0.0007]); pAML w/normal MC (OS [10.8 months vs NR, p=0.007], RFS [14.2 months vs NR, p=.04], EFS [7.1 months vs NR, p=0.009]); and pAML/sAML with abnormal MC (OS [4.6 months vs 8.5, p=0.04], EFS [2.5 vs 12.4 months,p=0.05]). Of key importance, the presence of copy-neutral loss of heterozygosity (LOH) also translates to worse outcomes (OS [4.2 vs 15.1 months, p=0.0018], EFS [2.6 vs 8.6 months, p=0.007]), a finding comparable to gains or traditional LOH. SNP-A also improved the ability to predict outcomes in both mutant and wild types (WT) based on Flt-3 ITD and NPM-1 mutation status, with inferior survival in patients with new defects detected by SNP-A (Flt-3 ITD mutant [OS: 8 months vs NR, p=0.0011; LFS: 14.2 months vs NR, p=0.04]; Flt-3 ITD WT [8.5 months vs NR, p=0.024]; NPM-1 mutant [OS: 12.1 months vs NR, p=0.03; LFS: 12.3 months vs NR, p=0.05]; and NPM-1 WT [OS: 5.7 months vs 15.1 months, p=0.04]). Multivariate analysis using the Cox proportional hazard method showed that the presence or absence of SNP-A defects is an independent predictive factor for OS (p=0.0076) and EFS (p=0.007). In conclusion, SNP-A improves the cytogenetic detection rate of MC. The detection of new chromosomal lesions, particularly copy-neutral LOH such as UPD, provides additional informative data to MC that is prognostically significant.