A nomogram based on CT radiomics and clinical features for predicting the metastatic risk of ALK rearrangement-positive lung adenocarcinoma

Objectives This study aimed to develop and validate a nomogram that integrates computed tomography (CT)-derived radiomics features with clinical characteristics to predict metastatic risk in patients with anaplastic lymphoma kinase (ALK) rearrangement-positive lung adenocarcinoma. The proposed model seeks to support individualized clinical decision-making in both preoperative and postoperative settings. Methods A retrospective cohort of 157 patients with confirmed ALK-positive lung adenocarcinoma was analyzed, comprising 85 patients with metastases and 72 without. Patients were randomly assigned to a training set (80%) and a validation set (20%). Radiomics features were extracted from pre-treatment CT scans and, together with clinical data, were incorporated into a RandomForest-based nomogram. Model performance was assessed using receiver operating characteristic (ROC) curve analysis, while clinical utility was evaluated through decision curve analysis (DCA). Model interpretability was enhanced using SHapley Additive exPlanations (SHAP). Results The combined radiomics-clinical nomogram demonstrated excellent predictive performance, achieving an area under the ROC curve (AUC) of 0.953 in the training set and 0.945 in the validation set. Its performance was comparable to the radiomics model (AUCs: 0.935 and 0.933; DeLong test P = 0.205 and 0.684) but significantly superior to the clinical model (AUCs: 0.922 and 0.857; DeLong test P = 0.029 and 0.045). Decision curve analysis confirmed the added clinical value of the nomogram across a range of threshold probabilities. SHAP analysis identified key radiomics features contributing to model predictions, improving transparency and interpretability. Conclusion The CT radiomics–based nomogram integrating clinical parameters offers a robust and interpretable tool for predicting metastatic risk in ALK-positive lung adenocarcinoma. Its application may facilitate more accurate risk stratification and enable personalized treatment planning, ultimately improving patient outcomes.