AUTHOR=Chen Xianyu , Tang Yongsheng , Wu Donghao , Li Ruixi , Lin Zhiqun , Zhou Xuhui , Wang Hezhen , Zhai Hang , Xu Junming , Shi Xianjie , Zhang Guangquan 

TITLE=From imaging to clinical outcome: dual-region CT radiomics predicting FOXM1 expression and prognosis in hepatocellular carcinoma

JOURNAL=Frontiers in Oncology

VOLUME=Volume 13 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/oncology/articles/10.3389/fonc.2023.1278467

DOI=10.3389/fonc.2023.1278467

ISSN=2234-943X

ABSTRACT=Background:
Hepatocellular carcinoma remains a significant global health challenge. Traditional prognostic indicators for HCC often fall short in providing comprehensive insights for individualized treatment. The integration of genomics and radiomics offers a promising avenue for enhancing the precision of HCC diagnosis and prognosis.
Methods:
From TCGA database, we categorized mRNA of HCC patients by Forkhead Box M1 (FOXM1) expression and performed univariate and multivariate studies to pinpoint autonomous HCC risk factors. The connection between FOXM1 and immune cells was evaluated using the CIBERSORTx database. The functions of FOXM1 were investigated through GO and KEGG analyses. After filtering through TCGA and TCIA  database, we employed dual-region CT radiomics technology to noninvasively predict the mRNA expression of FOXM1 in HCC tissues. Radiomic features were extracted from both tumoral and peritumoral regions, and a radiomics score (RS) was derived.A radiomics nomogram was developed by incorporating RS and clinical variables from the TCGA database. The models' discriminative abilities were assessed using metrics such as the AUC of the ROC curves and PR curves.
Results:
Our findings emphasized the overexpression of FOXM1 as a determinant of poor prognosis in HCC and illustrated its impact on immune cell infiltration. After selecting arterial phase CT, we chose 7 whole-tumor features and 3 features covering both the tumor and its surroundings to create WT and WP models for FOXM1 prediction. The WT model showed strong predictive capabilities for FOXM1 expression by PR curve. In our study, the RS was derived from whole-tumor regions on CT images. The RS was significantly associated with FOXM1 expression, with an AUC of 0.918 in the training cohort. Furthermore, the RS was integrated with clinical variables to develop a nomogram, which demonstrated good calibration and discrimination in predicting 12-, 36-, and 60-month survival probabilities. 
Conclusion:
This study highlights the potential of integrating FOXM1 expression and radiomics in understanding HCC's complexity. Our approach offers a new perspective in utilizing radiomics for non-invasive tumor characterization and suggests its potential in providing insights into molecular profiles. Further research is needed to validate these findings and explore their clinical implications in HCC management.