Publications

Identifying predictors of survival in patients with leukemia using single-cell mass cytometry and machine learning

Kleftogiannis et al. bioRxiv (2022)
A machine learning framework using mass cytometry data identifies predictive biomarkers of survival for leukemia patients
A machine learning framework using mass cytometry data identifies predictive biomarkers of survival for leukemia patients

Acute myeloid leukemia is a malignant blood cancer with many different subtypes. This study led by Dr. Dimitrios Kleftogiannis with senior authors Dr. Bjørn Tore Gjertsen and Dr.  Inge Jonassen at the University of Bergen combined mass cytometry and machine learning to explore signaling interactions in the blood cells of 45 leukemia patients for survival prediction.

The researchers selected 16 surface markers to define eight major blood cell types, and 14 intracellular markers to measure intracellular signaling networks. They found that patients with higher pP38-pSTAT5 and pCREB-pSTAT5 scores in the multipotent progenitor-like leukemia cells had a better chance of living longer. The European LeukemiaNet risk classification is widely used to predict survival outcomes in patients with acute myeloid leukemia. Remarkably, among 13 patients classified as “favorable” by the European LeukemiaNet, those with high pCREB-pSTAT5 and pP38-pSTAT5 scores had the better survival prognosis than those with low scores. 

This study confirms that mass cytometry profiling can improve survival prediction in acute myeloid leukemia when combined with the current classification system. The authors believe that machine learning modeling of mass cytometry data could serve as a framework for precision oncology.

Reference citation: Kleftogiannnis D, Tislevoll BS, Hellesøy M, Gullaksen S, van der Meer N, Griessinger E, Motzfeldt IKF, Fagerholt O, Lenartova A, Fløisand Y, Schuringa JJ, Gjertsen BT, Jonassen I. (2022).
bioRxiv 2022.08.13.503587; doi: 10.1101/2022.08.13.503587

You Might Also Like