A three-gene expression-based risk score can refine the European LeukemiaNet AML classification. Issue 1 (December 2016)

• Record Type: Journal Article
• Authors: Wilop, Stefan; Chou, Wen-Chien; Jost, Edgar; Crysandt, Martina; Panse, Jens; Chuang, Ming-Kai; Brümmendorf, Tim; Wagner, Wolfgang; Tien, Hwei-Fang; Kharabi Masouleh, Behzad
• Is Part Of: Journal of hematology & oncology. Volume 9:Issue 1(2016)
• Page Start: 1
• Page End: 10
• Publication Date: 2016-12
• Subjects: Hematology -- Periodicals; Oncology -- Periodicals; Blood -- Diseases -- Periodicals; 616.994
• DOI: 10.1186/s13045-016-0308-8
• Languages: English
• ISSNs: 1756-8722

Abstract

Abstract Background Risk stratification based on cytogenetics of acute myeloid leukemia (AML) remains imprecise. The introduction of novel genetic and epigenetic markers has helped to close this gap and increased the specificity of risk stratification, although most studies have been conducted in specific AML subpopulations. In order to overcome this limitation, we used a genome-wide approach in multiple AML populations to develop a robust prediction model for AML survival. Methods We conducted a genome-wide expression analysis of two data sets from AML patients enrolled into the AMLCG-1999 trial and from the Tumor Cancer Genome Atlas (TCGA) to develop a prognostic score to refine current risk classification and performed a validation on two data sets of the National Taiwan University Hospital (NTUH) and an independent AMLCG cohort. Results In our training set, using a stringent multi-step approach, we identified a small three-gene prognostic scoring system, named Tri-AML score (TriAS) which highly correlated with overall survival (OS). Multivariate analysis revealed TriAS to be an independent prognostic factor in all tested training and additional validation sets, even including age, current cytogenetic-based risk stratification, and three other recently developed expression-based scoring models for AML. Conclusions The Tri-AML score allows robust and clinically practical risk stratification for the outcome of AML patients. TriAS substantially refined current ELN risk stratification assigning 44.5 % of the patients into a different risk category.