Resumen:
Introducción: Se desarrolló un sistema de puntuación por ultrasonido (US) combinando siete descriptores ultrasonográficos para predecir nódulos linfáticos (LNs) metastásicos. Los objetivos de este estudio fueron (1) validar el Sistema de Reporte y Datos de Nódulos Cervicales por Ultrasonido (UNN-RADS) para características ultrasonográficas sospechosas de metástasis en LNs en pacientes que se someten a seguimiento por cáncer de tiroides y (2) evaluar el acuerdo interobservador. Materiales y Métodos: Se evaluaron pacientes en seguimiento por cáncer de tiroides mediante US y biopsia por aspiración con aguja fina (FNAB) guiada por US de los LNs. Se evaluó la puntuación ponderada de UNN-RADS de siete descriptores ultrasonográficos, a saber, forma, margen, ecogenicidad, hilio ecogénico, degeneración quística, calcificación, y patrón vascular intranodal para LNs malignos sospechosos. Las categorías de UNN-RADS se asignaron de acuerdo con la puntuación total. Se determinó el valor de corte óptimo y el rendimiento diagnóstico de la puntuación ponderada para predecir LNs malignos utilizando una curva de características operativas del receptor (ROC). Se utilizó el kappa de Cohen para calcular el acuerdo interobservador. Resultados: Incluimos 99 LNs de 99 pacientes que se sometieron a seguimiento por cáncer de tiroides, de los cuales 46 (46.5%) tenían LNs metastásicos y 53 (53.5%) LNs eran benignos. Los LNs metastásicos fueron de las categorías UNN-RADS 3 (22/46), 4 (15/46), y 5 (9/46), mientras que los LNs en las categorías UNN-RADS 1 y 2 eran benignos. Un corte de 6 puntos, correspondiente a la categoría UNN-RADS 3, tuvo una sensibilidad del 100% para predecir LNs malignos. La curva ROC mostró un AUC de 0.893 (IC del 95%, 81.5–94.6). El acuerdo interobservador entre los dos radiólogos fue bueno (κ = 0.71, IC del 95% 0.573–0.798). Conclusión: UNN-RADS es una herramienta confiable para predecir metástasis en LNs en pacientes con cáncer de tiroides durante el seguimiento. Las múltiples características en este sistema de puntuación, que pondera los hallazgos de US sospechosos de malignidad, pueden ser más precisas que una sola característica.

Abstract:
Introduction: An ultrasound (US) scoring system was developed by combining seven ultrasonographic descriptors for predicting metastatic lymph nodes (LNs). The aims of this study were (1) to validate the Ultrasound Neck Node Reporting and Data System (UNN-RADS) for suspicious ultrasonographic features of LN metastasis in patients undergoing thyroid cancer follow-up and (2) to evaluate the interobserver agreement. Materials and Methods: Patients undergoing thyroid cancer follow-up were evaluated with US and US-guided fine-needle aspiration biopsy (FNAB) of the LNs. The weighted UNN-RADS score of seven ultrasonographic descriptors, namely, shape, margin, echogenicity, echogenic hilum, cystic degeneration, calcification, and intranodal vascular pattern for suspicious malignant LN was evaluated. UNN-RADS categories were assigned according to the total score. The optimal cut-off value and diagnostic performance of the weighted score for predicting malignant LN were determined using a receiver operating characteristic (ROC) curve. Cohen’s kappa was used to calculate the interobserver agreement. Results: We included 99 LNs from 99 patients who underwent thyroid cancer follow-up, of which 46 (46.5%) had metastatic LNs and 53 (53.5%) LNs were benign. Metastatic LNs were UNN-RADS categories 3 (22/46), 4 (15/46), and 5 (9/46), while LNs in UNN-RADS categories 1 and 2 were benign. A cut-off of 6 points, corresponding to UNN-RADS category 3, had a sensitivity of 100% for predicting malignant LNs. The ROC curve showed an AUC of 0.893 (95% CI, 81.5–94.6). The interobserver agreement between the two radiologists was good (κ = 0.71, 95% CI 0.573–0.798). Conclusion: UNN-RADS is a reliable tool for predicting LN metastases in thyroid cancer patients during follow-up. The multiple features in this scoring system, which weighs US findings suspicious of malignancy, may be more accurate than a single feature.

Abstract:
Introduction: An ultrasound (US) scoring system was developed by combining seven ultrasonographic descriptors for predicting metastatic lymph nodes (LNs). The aims of this study were (1) to validate the Ultrasound Neck Node Reporting and Data System (UNN-RADS) for suspicious ultrasonographic features of LN metastasis in patients undergoing thyroid cancer follow-up and (2) to evaluate the interobserver agreement. Materials and Methods: Patients undergoing thyroid cancer follow-up were evaluated with US and US-guided fine-needle aspiration biopsy (FNAB) of the LNs. The weighted UNN-RADS score of seven ultrasonographic descriptors, namely, shape, margin, echogenicity, echogenic hilum, cystic degeneration, calcification, and intranodal vascular pattern for suspicious malignant LN was evaluated. UNN-RADS categories were assigned according to the total score. The optimal cut-off value and diagnostic performance of the weighted score for predicting malignant LN were determined using a receiver operating characteristic (ROC) curve. Cohen’s kappa was used to calculate the interobserver agreement. Results: We included 99 LNs from 99 patients who underwent thyroid cancer follow-up, of which 46 (46.5%) had metastatic LNs and 53 (53.5%) LNs were benign. Metastatic LNs were UNN-RADS categories 3 (22/46), 4 (15/46), and 5 (9/46), while LNs in UNN-RADS categories 1 and 2 were benign. A cut-off of 6 points, corresponding to UNN-RADS category 3, had a sensitivity of 100% for predicting malignant LNs. The ROC curve showed an AUC of 0.893 (95% CI, 81.5–94.6). The interobserver agreement between the two radiologists was good (κ = 0.71, 95% CI 0.573–0.798). Conclusion: UNN-RADS is a reliable tool for predicting LN metastases in thyroid cancer patients during follow-up. The multiple features in this scoring system, which weighs US findings suspicious of malignancy, may be more accurate than a single feature.