Construction of Two Independent RAB Family-Based Scoring Systems Based on Machine Learning Algorithms and Definition of RAB13 as a Novel Therapeutic Target for Hepatocellular Carcinoma

Abstract

Hepatocellular carcinoma (HCC) remains a global health challenge with a low early diagnosis rate and high mortality. The Rab GTPase (RAB) family plays an essential role in the occurrence and progression of HCC. Nonetheless, a comprehensive and systematic investigation of the RAB family has yet to be performed in HCC. We comprehensively assessed the expression landscape and prognostic significance of the RAB family in HCC and systematically correlated these RAB family genes with tumor microenvironment (TME) characteristics. Then, three RAB subtypes with distinct TME characteristics were determined. Using a machine learning algorithm, we further established a RAB score to quantify TME features and immune responses of individual tumors. Moreover, to better evaluate patient prognosis, we established a RAB risk score as an independent prognostic factor for patients with HCC. The risk models were validated in independent HCC cohorts and distinct HCC subgroups, and their complementary advantages guided clinical practice. Furthermore, we further confirmed that the knockdown of RAB13, a pivotal gene in risk models, suppressed HCC cell proliferation and metastasis by inhibiting the PI3K/AKT signaling pathway, CDK1/CDK4 expression, and epithelial-mesenchymal transition. In addition, RAB13 inhibited the activation of JAK2/STAT3 signaling and the expression of IRF1/IRF4. More importantly, we confirmed that RAB13 knockdown enhanced GPX4-dependent ferroptosis vulnerability, highlighting RAB13 as a potential therapeutic target. Overall, this work revealed that the RAB family played an integral role in forming HCC heterogeneity and complexity. RAB family-based integrative analysis contributed to enhancing our understanding of the TME and guided more effective immunotherapy and prognostic evaluation.