Improved outcome prediction of oropharyngeal cancer by combining clinical and MRI features in machine learning models

Objectives New markers are required to predict chemoradiation response in oropharyngeal squamous cell carcinoma (OPSCC) patients. This study evaluated the ability of magnetic resonance (MR) radiomics locoregional control (LRC) and overall survival (OS) after aimed determine whether this has added value traditional clinical outcome predictors. Methods 177 OPSCC patients were eligible for study. Radiomic features extracted from primary tumor region T1-weighted postcontrast MRI acquired before chemoradiation. Logistic regression models created using either variables (clinical model), radiomic (radiomic model) or combined (combined LRC OS 2-years posttreatment. Model performance was area under curve (AUC), 95 % confidence intervals calculated 500 iterations bootstrap. All analyses performed total population Human papillomavirus (HPV) negative subgroup. Results A model predicted treatment with a higher AUC (LRC: 0.745 [0.734–0.757], OS: 0.744 [0.735–0.753]) than 0.607 [0.594-0.620], 0.708 [0.697–0.719]). Performance comparable (AUC: 0.740 [0.729–0.750]), but not prediction 0.654 [0.646–0.662]). In HPV patients, all sufficient AUCs ranging 0.587 0.660 0.559 0.600 prediction. Conclusion Predictive that include derived MR images better based on only variables. perform OS.