AIM: To determine the accuracy of contrast-enhanced computed tomography (CECT) for nodal extracapsular spread (ECS) and identify predictive radiological signs and clinicopathological features for ECS in unknown-primary head and neck squamous cell cancer (UPHNSCC). MATERIALS AND METHODS: The CECT imaging of patients who underwent primary neck dissection for UPHNSCC during 2011–2015 was analysed. The largest pathological-looking node at each radiologically involved level was evaluated in consensus by two head and neck radiologists. Parameters included longest diameter, margin sharpness, haziness in adjacent fat, necrosis, and loss of fat plane with adjacent structures. Independent assessment was also made regarding the presence/absence of ECS. Findings and clinicopathological parameters were correlated with histopathology. RESULTS: Thirty-one patients with 39 neck levels had metastatic nodal involvement determined on CECT. Confirmed ECS was found at 26 levels in 23 patients. Sensitivity of radiological assessment for ECS by nodal level was 81–85% (95% confidence interval [CI]=65–93%) and specificity 46–54% (95% CI=19−81%); kappa 0.87. On univariate analysis based on the largest involved node per patient, longest diameter being ≥30 mm (p=0.007), haziness in adjacent fat (p=0.023), increasing age (p=0.006), and more advanced pathological nodal status (p=0.027) were statistically significantly associated with ECS. Haziness and increasing age were independent predictors on multivariate analysis (odds ratio [OR]=26.4 and 1.24). CONCLUSION: Expert assessment of ECS on CECT had good sensitivity with excellent interobserver agreement. A longest nodal diameter of ≥30 mm, haziness in the surrounding fat on CECT, advanced pathological nodal status, and advancing patient age were significantly associated with ECS in UPHNSCC patients, findings not previously reported.
|Number of pages||6|
|Early online date||1 Nov 2019|
|Publication status||Published - 31 Jan 2020|
- extracapsular nodal spread
- sqamous cell carcinoma
- contrast enhance computed tomography