Purpose: To determine the most accurate criterion for predicting malignancy of small cervical lymph nodes with gray-scale ultrasound (GSUS) and power Doppler ultrasound (PDUS).
Material and methods: Findings of 69 pathologically verified cervical nodes (38 benign, 31 malignant) in 57 patients without wide echogenic hilum on GSUS that measured less than 10 mm in minimal axial diameter were prospectively studied. Minimal and maximal axial diameters, ratios of minimal to maximal axial diameters, and presence or absence of calcification or necrosis of the nodes were assessed. On PDUS, vascularity in the node was classified into 4 pattern groups. A logistic model was used to evaluate the significant factors for predicting malignancy.
Results: The logistic model revealed that the minimal axial diameter and vascularity patterns were the only significant factors for malignancy. Using the minimal axial diameter, a node larger than 8 mm showed the highest accuracy (73%) with 45% sensitivity and 93% specificity. Of the vascularity patterns, spotted or peripheral pattern had the highest accuracy (80%) with 61% sensitivity and 93% specificity. A combined criterion of the minimal axial diameter larger than 8 mm and spotted or peripheral pattern increased the accuracy to 82% and sensitivity to 77% but specificity decreased to 86%.
Conclusion: A combined criterion of minimal axial diameter and vascular patterns was most accurate for diagnosing occult metastatic lymph nodes in the neck.