Predicting the clinical prognosis of non-small cell lung cancer patients by predicting ALOX5 expression: a radiomics model.

BACKGROUND

Arachidonic acid 5-lipoxygenase (ALOX5) may play an important role in non-small cell lung cancer (NSCLC) progression and treatment and may be a potential prognostic biomarker for NSCLC. This study aimed to predict the clinical prognosis of NSCLC patients by predicting ALOX5 expression using a radiomics model.

METHODS

Clinical and transcriptomic data of NSCLC patients were obtained from Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases and used for survival analysis (Kaplan-Meier survival curves: univariate and multivariate factors, Cox regression analysis, subgroup analysis and interaction test), correlation analysis of tumor clinical characteristics and immune cell abundance, and differential analysis of ferroptosis-related genes to evaluate the prognostic value of ALOX5. Contrast-enhanced computed tomography (CECT) scans of NSCLC patients from The Cancer Imaging Archive (TCIA) database were used to extract radiomics features to establish two radiomics models [logistic regression (LR) and Support Vector Machine (SVM) models]. Receiver operating characteristic (ROC), calibration, and decision curves were used to evaluate the two models, and the radiomics score (RS) of the model with the best prediction performance was selected to establish the Cox model for predicting NSCLC prognosis. A nomogram was used to visualize the prediction model, and its efficacy was evaluated and verified.

RESULTS

The prognostic value analysis of ALOX5 showed that high ALOX5 expression was a protective factor for overall survival (OS) of NSCLC patients, and it negatively correlated with histology (P<0.001). Overall, 107 features were obtained from CECT images of NSCLC patients, and 8 optimal features were selected. The LR [area under the curve (AUC) =0.783] and SVM (AUC =0.763) models with good performance and clinical benefit were established using the LR and SVM algorithms, respectively. The RS output by the LR model strongly correlated with ALOX5 expression (P<0.05).

CONCLUSIONS

The findings suggest that evaluating ALOX5 expression using a radiomics model to predict the clinical prognosis of NSCLC patients could have potential clinical applications.