Abstract
BACKGROUND: T staging is closely linked to the prognosis of patients with nasopharyngeal carcinoma (NPC), and dual-energy computed tomography (DECT) has a lot of promise for determining the degree of tumor invasion in head and neck cancers. This study aims to evaluate the value of quantitative parameters of DECT in distinguishing between T3 and T4 stages of locally advanced nasopharyngeal carcinoma (LA-NPC) and to compare the diagnostic performance of DECT and magnetic resonance imaging (MRI). METHODS: This prospective study enrolled 136 patients with histologically confirmed LA-NPC. This cohort was divided into T3 and T4 stage, and the performance of DECT for staging was compared to that of MRI. Subsequently, logistic regression analysis was performed on the quantitative parameters of DECT, and a combined model was obtained. The receiver operating characteristic (ROC) curves were used to assess the diagnostic performances of the individual DECT-derived parameter and the combined model in distinguishing T3 from T4 stage LA-NPC. RESULTS: Multivariate logistical analysis showed the DECT-derived parameters, such as the normalized iodine content in the venous phase (V-NIC) and the relative electron density related to water in the arterial phase (A-Rho) were independent influencing factors for distinguishing T3 and T4 stages of LA-NPC. The area under the curve (AUC) values for the V-NIC, A-Rho, and the combined model based on V-NIC and A-Rho parameters were 0.64, 0.81, and 0.84, with sensitivity of 46.27%, 89.55%, and 77.61% and specificity of 82.61%, 60.87%, and 80.43%, respectively. The qualitative diagnostic performance of DECT parameters (AUC of 0.92) was higher than the MRI (AUC of 0.88) for differentiating the T3 from T4 stage in patients with LA-NPC (P=0.19), with 124 patients (124/136=91.18%) correctly diagnosed by DECT and 118 patients (118/136=86.76%) correctly diagnosed by MRI. CONCLUSIONS: The T3 and T4 stages of LA-NPC can be distinguished more accurately by the morphology of DECT than by MRI, although not reaching statistical significance. The model combining V-NIC and A-Rho parameters from DECT enhanced diagnostic performance.