A Novel Machine Learning-Derived Radiomic Signature Predictive of Nasopharyngeal Necrosis to Guide Re-Radiotherapy for Recurrent Nasopharyngeal Carcinoma: A Multicentre Study

Background: Accurate pre-treatment assessment of the risk of post-radiation nasopharyngeal necrosis (PRNN) is crucial to patient selection and tailoring re-radiotherapy regimens. We aimed to develop a radiomic signature for the pre-treatment prediction of PRNN to guide salvage re-radiotherapy in patients with locally recurrent nasopharyngeal carcinoma (LRNPC).Methods: This multicentre study included 761 patients with LRNPC who were treated with curative re-radiotherapy at four centres in China and divided them into training, internal validation, and external validation cohorts. We built a machine learning (random forest) radiomic signature based on the pre-treatment multiparametric magnetic resonance images for predicting PRNN following re-radiotherapy. We comprehensively assessed the performance of the radiomic signature. Gene set enrichment analyses were conducted to identify the associated biological processes.Findings: The radiomic signature showed optimal discrimination of 1-year PRNN in the training (area under the curve (AUC) 0.722, 95% confidence interval (CI) 0.676-0.765), internal validation (AUC 0.713, 95% CI 0.653-0.772), and external validation (AUC 0.756, 95% CI 0.673-0.838) cohorts. Stratified by a cutoff radiomics score of 0.735, patients with high-risk signature had higher incidences of PRNN than patients with low-risk signature in all cohorts (1-year PRNN rates 42.2%-62.5% vs. 16.3%-18.8%, P<0.001). The signature significantly outperformed the clinical model (P<0.05) and was generalizable across different centres, imaging parameters and patient subgroups. Radiogenomics analyses revealed associations between the radiomic signature and signaling pathways involved in tissue fibrosis and vascularity.Interpretation: We present a radiomic signature for the individualized risk assessment of PRNN following re-radiotherapy, which may serve as a noninvasive radio-biomarker of radiation injury-associated processes and a useful clinical tool to personalize treatment recommendations for patients with LANPC.Funding Information: This study was funded by grants from the National Key R&D Program of China (2017YFC0908500, 2017YFC1309003), the National Natural Science Foundation of China (No. 81425018, No. 81672868, No.81802775,No. 82073003, No.82002852, No. 82003267, No. 82022036, No. 91959130, No. 81971776, No. 81771924, No. 62027901, No. 81930053), the Sci-Tech Project Foundation of Guangzhou City (201707020039), the Sun Yat-sen University Clinical Research 5010 Program (No. 2019023), the Special Support Plan of Guangdong Province (No. 2014TX01R145), the Natural Science Foundation of Guangdong Province (No.2017A030312003, No.2018A0303131004), the Natural Science Foundation of Guangdong Province for Distinguished Young Scholar (No. 2018B030306001), the Sci-Tech Project Foundation of Guangdong Province (No. 2014A020212103), the Health & Medical Collaborative Innovation Project of Guangzhou City (No. 201400000001, No.201803040003), the Planned Science and Technology Project of Guangdong Province (2019B020230002), the National Science & Technology Pillar Program during the Twelfth Five-year Plan Period (No. 2014BAI09B10), Natural Science Foundation of Guangdong Province (2017A030312003), the Key Youth Teacher Cultivating Program of Sun Yat-sen University (20ykzd24), the Fundamental Research Funds for the Central Universities, the Beijing Natural Science Foundation (L182061), the Strategic Priority Research Program of Chinese Academy of Sciences (XDB 38040200), and the Youth Innovation Promotion Association CAS (2017175).Declaration of Interests: The authors declared no conflict of interest.Ethics Approval Statement: This study was approved by the clinical research committee of the study centres, and written informed consent was retrieved from all included patients.