Abstract 4885: Integrating mutational profiles and transcriptional data with ASTUTE to elucidate the key molecular functions involved in the pathogenesis of cancer

Abstract Cancer is a highly heterogeneous disease characterized by genomic and phenotypic changes that differ among tumor subtypes. A comprehensive understanding of the molecular heterogeneity is crucial for identifying molecular biomarkers and developing targeted therapies. While recent years have seen significant progress in genomic studies and advancements in next-generation sequencing technologies, elucidate the impact of individual genomic alterations on the transcriptome landscape of cancer cells remains essential. This knowledge is crucial for developing personalized therapies. To integrate mutational profiles and transcriptional data, we have developed a novel computational framework named ASTUTE (Association of SomaTic mUtaTions to Expression). ASTUTE establishes associations between somatic mutations and gene expression profiles, quantifying the results as fold changes and comparing the effects of individual mutations. In particular, ASTUTE leverages LASSO regularized regression for feature selection, identifying the most relevant mutated genes that exhibit a strong association with expression levels. Our goal is to create a valuable resource for identifying diagnostic markers and advancing the development of targeted therapies in cancer. To this end, we applied ASTUTE to two published bulk RNA-seq datasets of adult acute myeloid leukemia (AML) patients from the Beat AML program and the TCGA study to verify the capacity of our approach to identify genes whose expression correlates with the presence of specific somatic mutations. The first dataset comprises 585 samples, while the second one includes 173 specimens, both of them contain mutations and RNA-seq data. We executed ASTUTE on the two datasets independently, considering all genes for RNA-seq data and the alterations in the top 10 most mutated genes. We considered the associations consistently discovered in both cohorts. Our analysis revealed a strong association between NPM1 mutations and the expression of HOX genes. Specifically, HOXB6, HOXB5, HOXB3, HOXA5, HOXB2, and HOXA10 exhibited a log2 fold change higher than 1.5. The upregulation of these genes has been reported in association with NPM1 mutations in AML adult patients. Additionally, ASTUTE identified HOXA6, HOXA7, HOXA4, and HOXA9, whose expression has been described to be upregulated in the presence of NPM1 mutations in AML pediatric samples. Finally, we identified a robust relationship between NPM1 mutations and the expression of PBX3, MEIS, and ITM2A. The first two genes are upregulated in the presence of NPM1 mutations, while the third one is downregulated. Our analysis showcases ASTUTE's capability to identify potential markers and underscores the possibility to apply ASTUTE to different tumors datasets, with the aim to better characterize cancer heterogeneity and develop targeted therapies. Citation Format: Valentina Crippa, Diletta Fontana, Ivan Civettini, Luca Mologni, Rocco Piazza, Carlo Gambacorti-Passerini, Daniele Ramazzotti. Integrating mutational profiles and transcriptional data with ASTUTE to elucidate the key molecular functions involved in the pathogenesis of cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4885.