MultiClass Machine Learning-based Identification of Anoikis-related Genes Across Three Adult T-cell Leukemia/Lymphoma Subtypes

Abstract Adult T-cell leukemia/lymphoma (ATLL) is a type of cancer that originates from T-cells infected with the human T-cell lymphotropic virus type 1 (HTLV-1). Anoikis is a programmed cell death process that occurs when the contact between cells or the extracellular matrix is lost. We aim to identify the specific anoikis-related classifier genes for three ATLL subtypes, which could provide valuable insights into the molecular mechanisms underlying the disease's progression and potential targets for intervention. We conducted an analysis of differentially expressed anoikis genes (DEAGs) for identifying those associated with each subtype. Subsequently, we utilized multiclass support vector machine and logistic regression algorithms to recognize specific classifier anoikis-related genes distinguishing each ATLL subtype. The results revealed the activation of several cancer- and anoikis-related pathways. Moreover, specific potential biomarkers were pointed out for each ATLL subtype: acute, with S100A9 and MAOA; chronic, with IL10, CDH1, and CYP3A4; and smoldering with BCL2L1 and SNAI2. These anoikis-related genes play a role in regulating cell adhesion and survival signaling which are crucial for maintaining normal cellular homeostasis. The findings not only contribute to our understanding of ATLL progression but also offer potential targets for developing more effective therapeutic strategies and improving treatment outcomes for patients with different subtypes.