Differential diagnosis of cervical lymphadenopathy: Integration of postvascular phase of contrast-enhanced ultrasound and predictive nomogram model
Xiangmei Chen,Jiuyi Ma,Ying Fu,Fang Mei,Rui Tang,Heng Xue,Yuxuan Lin,Shumin Wang,Ligang Cui
DOI: https://doi.org/10.1016/j.ejso.2024.107981
IF: 4.037
2024-01-25
European Journal of Surgical Oncology
Abstract:Background Distinguishing benign from malignant cervical lymph nodes is critical yet challenging. This study evaluates the postvascular phase of contrast-enhanced ultrasound (CEUS) and develops a user-friendly nomogram integrating demographic, conventional ultrasound, and CEUS features for accurate differentiation. Methods We retrospectively analyzed 395 cervical lymph nodes from 395 patients between January 2020 and December 2022. The cohort was divided into training and validation sets using stratified random sampling. A predictive model, based on demographic, ultrasound, and CEUS features, was created and internally validated. Results The training set included 280 patients (130 benign, 150 malignant nodes) and the validation set 115 patients (46 benign, 69 malignant). Relative hypoenhancement in the postvascular phase emerged as a promising indicator for MLN, with sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of 96.7 %,52.3 %, 70.0 %, 93.2 %, and 76.1 %, respectively in the training set and 95.7 %, 52.2 %, 75.0 %, 88.9 %, and 74.8 % in the validation set. Age over 50 years, history of malignancy, short-axis diameter greater than 1.00 cm, focal hyperechogenicity, ill-defined borders, and centripetal perfusion were also identified as independent MLN indicators. The nomogram prediction model showed outstanding accuracy, with an area under the curve (AUC) of 0.922 (95 % CI: 0.892–0.953) in the training set and 0.914 (95 % CI: 0.864–0.963) in the validation set. Conclusion Relative hypoenhancement in the postvascular phase of CEUS, combined with demographics and ultrasound features, is effective for identifying MLNs. The developed prediction model, with a user-friendly nomogram, can facilitate clinical decision-making.
oncology,surgery