SNP-Array-Based Karyotyping Has Impact on Cytogenetic Diagnosis and Prognosis of Non-Core Binding Factor Primary and Secondary AML.

Abstract

Non-core binding factor AML has heterogeneous clinical phenotypes, likely due to various modifying genetic lesions (i.e. point mutations such as Flt3, c-kit, or and nucleophosmin). Using metaphase cytogenetics (MC), chromosomal (chr.) abnormalities are found in only 50% of newly diagnosed patients with primary AML (pAML) and secondary AML (sAML) arising from MDS or MPD. Previously, we have demonstrated that SNP arrays (SNP-A) can detect previously cryptic lesions (including uniparental disomy, UPD) and enhance the clinical value of MC in patients with MDS. Here, we hypothesize that SNP-A will improve cytogenetic analysis in AML as well. Our study included 79 healthy control marrows and 103 AML cases; 36 pAML (FAB M0=3, M1=10, M2=10, M4=6, M5=6, M7=1; mean age 53y) and 67 sAML (from MDS, N=40 and MPD/MDS, N=27; mean age 63y). Normal MC was present in 69% and 45% of pAML and sAML, respectively. First, we investigated technical aspects of SNP-A karyotyping. Dilution studies showed that SNP-A can detect clones spanning 25–50% as well as LOH calls >99% of the time as shown X chr. analysis in males. Repetitive/serial testing demonstrated congruent results and somatic derivation of randomly selected lesions was confirmed by microsatellite and SNP-A of non-clonal cells. Copy number variants (CNV) encountered in controls or described in public databases were excluded. Using SNP-A, new cytogenetic abnormalities were found in 52% (28% UPD) and 59% (33% UPD) of pAML and sAML with normal MC, respectively. Moreover, 80% and 88% of pAML and sAML with previously abnormal MC harbored lesions detected by SNP-A. Examples of microdeletions/duplications include regions harboring known leukemia susceptibility genes, such as AML1. Segmental UPD involved regions often affected by deletion, including 5q, 7q, and 11q among others. Results of SNP-A can help characterize recurrent or minimally shared lesions, map their location, or identify causative genes. However, clinical utility of this technology is best demonstrated by the impact of the new defects on survival and other clinical parameters. In both pAML and sAML patients, we found that those with both normal MC and normal SNP-A had a better overall survival (OS) and event-free survival (EFS) as compared to those showing normal MC but abnormal SNP-A. (pAML: OS: p=.04, 21 vs. 5mo; EFS: p=.03, 19 vs. 6mo; sAML: OS: p=.04, 15 vs. 4mo; EFS: p=.04, 10 vs. 4mo). A subset analysis of those sAML patients derived from MDS showed similar results (OS: p=.02, 20 vs. 4mo; EFS: p=.03, 16 vs. 6mo). Most significantly, new lesions detected by SNP-A in AML patients with previously abnormal MC corresponded to a worse prognosis (OS: p=.0004, 10 vs. 3mo). For frequently encountered lesions, we performed survival analysis. For example, the presence of UPD7q negatively affected clinical outcomes (5 patients with UPD7 had equally poor survival to patients with del7/7q, N=10). Subset analyses (e.g., AML with normal MC) also indicated that chr. lesions detected by SNP-A impact stratification schemes independent of known risk factors such as Flt3 mutational status. In summary, SNP-A karyotyping allows for detection of previously cryptic cytogenetic lesions that together with routine MC may aid not only in diagnosis but prognosis in patients with both pAML and sAML.