Identification of AKTIP as a biomarker for fibrolamellar carcinoma using WGCNA and machine learning

Fibrolamellar carcinoma (FLC) is a rare form of liver carcinoma with limited diagnostic and therapeutic options. In this study, we utilized the GSE57727 and E-MTAB-1503 datasets, downloaded from GEO and ArrayExpress, respectively, to explore hub genes for FLC diagnosis and potential therapeutic agents. Through the integration of multiple machine learning approaches and drug sensitivity databases, we identified AKTIP as a potential diagnostic biomarker for FLC. AKTIP exhibited markedly elevated expression in FLC compared to non-FLC, demonstrating superior diagnostic and prognostic performance over other FLC-specific biomarkers. Four compounds (PI-103, BVT-948, Digitoxigenin, and SB-218078) were identified as potential therapeutic agents targeting AKTIP. Molecular docking analysis revealed strong binding affinities of these compounds to AKTIP, and molecular dynamics simulations further validated the reliability and rationality of the molecular docking results. Pan-cancer analysis indicated that AKTIP expression varies across different tissues and is significantly associated with patient prognosis. qRT-PCR analysis confirmed that AKTIP mRNA levels were markedly overexpressed in normal liver epithelial cells compared to human hepatocellular carcinoma cell lines. In conclusion, AKTIP was successfully identified as a diagnostic and prognostic biomarker for FLC, and four compounds were proposed as potential therapeutic agents. This study uncovers new perspectives on diagnosing and managing of this rare type of liver carcinoma, offering promising avenues for future research and clinical applications.

The online version contains supplementary material available at 10.1007/s13205-025-04323-4.