The treatment responses and overall survival of patients with AML is known to be heterogeneous, influenced both by patient-specific as well as disease-specific factors, making AML a challenging disease for risk scoring and treatment recommendations.
A team led by Professor Tero Aittokallio from the Institute for Molecular Medicine Finland FIMM, University of Helsinki, has developed a novel prognostic gene expression signature and risk score, named IPRP. In their recent publication in Leukemia, they demonstrated that the IPRP score leads to robust and accurate survival prediction performance.
The robust statistical approach is based on relative differences across 10 gene pairs that effectively eliminates confounding variability across heterogeneous patient cohorts and transcriptomic data distributions.
The team established the prognostic gene-pair signature and the associated risk indicator using a total of 1327 AML patients from five independent patient cohorts. Despite the differences in the treatment protocols and genomic background of the patients, the robust IPRP signature showed consistent performance across the AML cohorts. Quite unexpectedly, the signature showed associations with survival differences and tumor-immune interactions also in multiple solid cancer types in a pan-cancer analysis of more than 10,000 patients.
“Our unique gene-pairing approach enables one to define a universal risk score for distinguishing between high- and low risk patients, regardless of a particular AML population or the transcriptomic technology used, making the signature widely applicable for any patient cohort”, explains Doctoral Researcher Weikaixin Kong from FIMM, the fist author of the study and the main developer of the method.
“Compared to the current standards for AML patient stratification, IPRP was shown to provide a significant added value for improved prognostic prediction of AML patients”, says Doctor Liye He, another lead author of the work.
Towards routine clinical applications, the IPRP score was implemented as an easy-to-use 4-factor risk predictor and a web-tool for AML risk scoring and therapy stratification.
The signature combines pyroptosis-related genes with immune-related genes, which was shown to improve the prognostic classification accuracy. Even though not originally designed for immunotherapy response prediction, the use of immune-related genes enables wider application of the score for the analysis of tumor-immune interactions in multiple cancer types, toward future immunotherapy stratification, as was demonstrated using follow-up analyses of total mutation burden, target expression of various immunotherapies, and immune infiltration analyses.
“Even if developed in AML patient cohorts, we expect the approach will have also a wider application for survival prediction and analysis of tumor immunity in multiple cancer types”, said Professor Tero Aittokallio.