3197 – RE-PURPOSING OF GENE SIGNATURES IN AML UNCOVERS NOVEL MOLECULAR SUBTYPES ASSOCIATED WITH ENERGETICS

Abstract

AML is a highly heterogeneous disease with great diversity in patient treatment response. This presents the need to better characterise the disease and its underpinning molecular subtypes. The re-use of published and validated prognostic predictive gene signatures present an invaluable in silico opportunity to uncover the biological mechanisms underpinning treatment response in AML. Analysis was carried out using a primary dataset (AML-OHSU) and a validation dataset (TCGA-LAML: 200 AML patient samples). Both datasets had been processed using Almac's claraT platform, a software solution which provides a comprehensive overview of tumour profiles via gene expression signatures. An automated analytical pipeline was developed using Consensus Clustering, a method that determines the number and membership of patient clusters within a dataset. Robustness was tested via bootstrapping. This pipeline was used to stratify patients using 451 gene signatures categorised by 10 different hallmarks of cancer. The automated clustering pipeline analysed a total of 1,314 stable clusters in the AML-OHSU dataset. Stable clusters were subsequently processed via log rank analysis identifying 134 stable clusters with significant differences (p-value < 0.05) in survival outcome. Here we found gene signatures representative of the Energetics hallmark to be one of the most frequently clustered throughout our results, ranking highest where K=3. A significant difference in overall survival probability (Log rank p-value: 0.033) was found between clusters. To validate energetics results from the AML-OHSU dataset, consensus clustering was again performed using gene signatures from the TCGA dataset that were representative of the energetics hallmark. A significant difference between overall survival probability (Log rank test p-value: 0.019) was again found between stable clusters AML is a highly heterogeneous disease with great diversity in patient treatment response. This presents the need to better characterise the disease and its underpinning molecular subtypes. The re-use of published and validated prognostic predictive gene signatures present an invaluable in silico opportunity to uncover the biological mechanisms underpinning treatment response in AML. Analysis was carried out using a primary dataset (AML-OHSU) and a validation dataset (TCGA-LAML: 200 AML patient samples). Both datasets had been processed using Almac's claraT platform, a software solution which provides a comprehensive overview of tumour profiles via gene expression signatures. An automated analytical pipeline was developed using Consensus Clustering, a method that determines the number and membership of patient clusters within a dataset. Robustness was tested via bootstrapping. This pipeline was used to stratify patients using 451 gene signatures categorised by 10 different hallmarks of cancer. The automated clustering pipeline analysed a total of 1,314 stable clusters in the AML-OHSU dataset. Stable clusters were subsequently processed via log rank analysis identifying 134 stable clusters with significant differences (p-value < 0.05) in survival outcome. Here we found gene signatures representative of the Energetics hallmark to be one of the most frequently clustered throughout our results, ranking highest where K=3. A significant difference in overall survival probability (Log rank p-value: 0.033) was found between clusters. To validate energetics results from the AML-OHSU dataset, consensus clustering was again performed using gene signatures from the TCGA dataset that were representative of the energetics hallmark. A significant difference between overall survival probability (Log rank test p-value: 0.019) was again found between stable clusters