919P - A prognostic model of nasopharyngeal carcinoma based on 18F-FDG PET-CT radiomic parameters and clinical characteristics of patients

Background

18F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography (18F-FDG PET-CT) is a non-invasive imaging modality. This study aimed to investigate the effects of radiometric parameters and clinical characteristics of 18F-FDG PET-CT on the prognosis of patients with nasopharyngeal carcinoma (NPC).

Methods

A total of 801 patients with metastatic NPC were enrolled between 2016 and 2020, and all participants underwent 18F-FDG PET/CT before treatment. We assessed the following 18F-FDG PET parameters: standardized uptake value (SUV), SUV-avg (average), SUV-peak (maximum), and total lesion glycolysis (TLG) of the primary tumor and regional lymph nodes. We used multivariate Cox proportional hazards models and multiple machine learning techniques to identify independent predictors of survival based on radiomic parameters and patient clinical characteristics. After selecting the features, we established prognosis models using a COX stepwise regression model. Finally, we performed internal validation, and a nomogram was created for NPC comprehensive diagnosis.

Results

In this study, we developed a stable and robust model incorporating five independent prognostic features based on COX regression. The area under the curve (AUC) of the model was 0.75 (95% CI = 0.52–0.97) at 1 year, 0.80 (95% CI = 0.71–0.89) at 3 years, and 0.86 (95% CI = 0.78–0.94) at 5 years. The AUC at 3 years and 5 years were calculated as 0.7276 and 0.7242, respectively. The model was validated using receiver operating characteristic (ROC), consistency index, and multivariate analysis, which showed high accuracy and consistent performance. Moreover, patients in the low-risk score group exhibited significantly better outcomes than those in the high-risk score group (P < 0.001).

Conclusions

Our findings suggest that age, SUV-Tmax, SUV-Tavg, SUV-Tpeak, and TLG-T are valuable parameters for predicting long-term survival in patients with NPC. The nomogram based on metabolic parameters combined with other variables showed good prognostic accuracy in predicting NPC prognosis.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.