Dual-region-based computed tomography radiomics analysis for the non-invasive prediction of telomerase reverse transcriptase status and clinical prognosis in liver cancer

Abstract

Abstract Background: Telomerase reverse transcriptase (TERT) can directly regulate various hallmarks of cancer. We aimed to estimate the prognostic value of TERT expression levels in patients with liver cancer and build a radiomics model that can predict the TERT expression levels using The Cancer Imaging Archive (TCIA) and The Cancer Genome Atlas (TCGA) databases. Methods: Preoperative CT images stored in TCIA with genomic data from TCGA were used for radiomics feature extraction and model construction. The radiomics features were extracted using least absolute shrinkage and selection operator regression analysis. A logistic regression algorithm was used to construct the model and to extract features based on whole tumor and whole tumor-peritumoral regions; a prognostic scoring system incorporating a radiomics signature based on the TERT expression levels was accepted for survival prediction. Results: TCGA data on 295 liver cancer cases (203 men; age <60 years, 142 and ≥60 years, 153 participants) were used for gene-based survival analysis. High TERT expression was an independent risk factor for overall survival (OS) deterioration, involved in immune cell infiltration and ferroptosis, and closely related to several signaling pathways. The 34 cases included in the radiomics model for predicting TERT expression levels achieved areas under the curve of 0.827 and 0.803 in the training and validation sets, respectively. The inclusion of clinical features and important imaging biomarkers can improve the model’s accuracy of OS estimation. Conclusion: Radiomics can predict the prognosis of patients with hepatocellular carcinoma by predicting TERT expression. CT-based radiomics can serve as a novel and effective tool for predicting prognosis in clinical settings.