A Combined Radiomics and Immunological Checkpoint Molecular Markers Based on Multi-Sequence Magnetic Resonance Imaging to Predict Long-Term Survival of Nasopharyngeal Carcinoma Patients
Abstract
Purpose
To develop and validate a combined model, based on multi-sequence magnetic resonance imaging (MRI), integrating immune checkpoint molecular markers and multiple independent prognostic factors, to predict overall survival (OS) in nasopharyngeal carcinoma (NPC) patients.
Methods
This retrospective study included 124 NPC patients who underwent MRI examinations. Using the least absolute shrinkage and selection operator (LASSO) regression within a Cox regression model, we identified the optimal Rad score by selecting 15 features from a total of 2,553 radiomics features derived from multi-sequence MRI (T1, T2, T1C). We then compared its discriminatory ability across training and validation cohorts. We developed an optimized nomogram incorporating the optimal Rad score, immune checkpoint molecular markers, and clinical risk factors to enhance predictive performance.
Results
Statistical analysis revealed statistically significant differences in certain clinical data (age, DM) and immunological molecular markers (CD4, CD4PD-1, CD8CTLA-4) (P<0.001). Machine learning model results indicated that ExtraTrees demonstrated optimal diagnostic performance. This suggests that Combined2 (multiple-sequence radiomics features combined with clinical risk factors and immune checkpoint markers) (accuracy 0.942) exhibited superior diagnostic efficacy compared to Combined1 (multiple-sequence radiomics features combined with clinical risk factors) (accuracy 0.895).
Conclusion
A combined model incorporating multi-sequence MRI radiomics and clinical risk factors (Combined1) can assess long-term prognosis in nasopharyngeal carcinoma patients. The integrated model (Combined2) combining multi-sequence radiomics, clinical risk factors, and immune checkpoint molecular markers further enhances predictive accuracy, providing a reference basis for early assessment of long-term prognosis and development of personalized treatment strategies in nasopharyngeal carcinoma patients.