Validation of risk stratification models in acute myeloid leukemia using sequencing-based molecular profiling

Area of Science:

• Hematology
• Oncology
• Genomics

Background:

• Accurate risk stratification for acute myeloid leukemia (AML) is essential for effective treatment planning.
• Existing AML risk classification models require systematic validation for clinical use.
• Molecular profiling, including gene-expression and somatic mutations, offers potential for improved AML prognostication.

Purpose of the Study:

• To independently validate existing molecular-based AML risk stratification models.
• To assess the added prognostic value of molecular data to current clinical risk classifications.
• To evaluate novel risk classification systems combining molecular and clinical data for AML patients.

Main Methods:

• Whole-transcriptome RNA-sequencing and deep DNA sequencing of 23 genes were performed on 274 AML patients (Clinseq-AML cohort).
• The Cancer Genome Atlas (TCGA)-AML study (N=142) served as an independent validation cohort.
• Six established molecular risk models and two revised systems integrating molecular and clinical data were evaluated.

Main Results:

• Five of six molecular risk models demonstrated significant differences in overall survival among cytogenetically normal AML patients in the Clinseq-AML cohort.
• Molecular-based risk classification systems added prognostic value to the European Leukemia Net (ELN) classification.
• Model performance and prognostic value varied across models and cohorts, emphasizing the need for validation.

Conclusions:

• Molecular-based risk models hold potential for improving AML patient stratification.
• Combining molecular and clinical data offers a promising approach for enhanced AML risk classification.
• Independent validation is critical to establish the efficacy and general applicability of AML risk models.