Hierarchy in Gene Expression is Predictive for Adult Acute Myeloid Leukemia

TL;DR

This study introduces a measure of gene network hierarchy in AML that correlates with relapse risk and could serve as a prognostic biomarker, revealing complex evolutionary dynamics.

Contribution

It defines a novel hierarchy measure in gene networks and demonstrates its predictive value for AML prognosis using patient gene expression data.

Findings

Hierarchy level correlates with relapse risk.

Gene network organization varies non-monotonically during progression.

Hierarchy at diagnosis predicts clinical outcomes.

Abstract

Cancer progresses with a change in the structure of the gene network in normal cells. We define a measure of organizational hierarchy in gene networks of affected cells in adult acute myeloid leukemia (AML) patients. With a retrospective cohort analysis based on the gene expression profiles of 116 acute myeloid leukemia patients, we find that the likelihood of future cancer relapse and the level of clinical risk are directly correlated with the level of organization in the cancer related gene network. We also explore the variation of the level of organization in the gene network with cancer progression. We find that this variation is non-monotonic, which implies the fitness landscape in the evolution of AML cancer cells is nontrivial. We further find that the hierarchy in gene expression at the time of diagnosis may be a useful biomarker in AML prognosis.