Abstract
BACKGROUND/OBJECTIVES: Tetraploidy (4n = 92 chromosomes) and near-tetraploidy (81-103 chromosomes) (T/NT) are uncommon cytogenetic events in MDS/AML (~1%). Abnormalities reported to be associated with T/NT MDS/AML include -5/del(5q), -7/del(7q), +8, and +21. However, other clinically relevant abnormalities likely remain "hidden" in long strings of ISCN cytogenetic nomenclature when evaluated visually. To date, no studies have had the statistical power and a computational method to identify novel recurrent abnormalities associated with the T/NT karyotype or overall survival (OS). METHODS: Using CytoGPS, a bioinformatic tool we developed, we converted karyotypes from a combined cohort of 75 T/NT MDS/AML cases from two institutions into a binary Loss-Gain-Fusion model, which is analyzable using computational methods. RESULTS: On univariate analyses, age as a continuous variable (p = 0.032), prior treatment (p = 0.011), and cohort (p = 0.025) were associated with OS; age ≥ 60 years (p = 0.316), gender (p = 0.916), karyotypic complexity (p = 0.175), time from diagnosis to T/NT karyotype identification (p = 0.419), and clone size (p = 0.316) had no effect. Univariate analyses of karyotypes demonstrated that -5, -16, -18, del(11)(p15.1p15.4), del(13)(q12.11q22.3), and +8 were associated with poorer OS (unadjusted p < 0.05). CONCLUSIONS: Using the results of univariate analyses to build multivariate models of OS, the best predictor of OS was the presence of any one of these six cytogenetic abnormalities.