Video Experimental Relacionado
Updated: Jun 16, 2026
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
Desarrollo de nomogramas PET/CT-clínicos para predecir metástasis de ganglios linfáticos en cáncer de pulmón primario
Xiaoyu Han1, Wentao Li1, Julie Lin2
1Department of Imaging Physics, The University of Texas MD Anderson Cancer Center, Houston, USA.
Objective:
Accurate staging of primary lung cancer is crucial for optimizing therapeutic strategies but remains challenging in clinical practice. We aimed to develop a nomogram incorporating clinical characteristics with CT and PET findings to predict lymph node metastasis (LNM) in primary lung cancer.
Materials And Methods:
We retrospectively analyzed patients with primary lung cancer and mediastinal and hilar LNs from a tertiary care cancer center. All patients underwent endobronchial ultrasound-guided transbronchial needle aspiration, with diagnostic chest CT and PET-CT. Cytological confirmation of transbronchial needle aspiration samples served as the gold standard for diagnosing LNM. We employed an LN-level modeling approach and constructed five models for independent prediction of LNM: (1) Clinical-CT-PET model, (2) Clinical-CT model, (3) PET model, (4) Clinical-PET model, and (5) CT-PET model. Their performance was further evaluated in the subgroup of LNs < 1 cm.
Results:
This study included 455 patients (mean age 70 ± 10 years; 55.4% male), predominantly adenocarcinoma (62.0%). Most (68.1%) were stage III-IV. In total, 1391 lymph nodes (1112 training, 279 testing) were analyzed to develop and validate the nomogram. The Clinical-CT-PET model achieved the best diagnostic performance, with AUCs of 0.883 (training cohort) and 0.877 (test cohort), sensitivities of 79.5% and 80.0%, and specificities of 87.1% and 86.9%, respectively. For small LNs, it showed higher AUC (0.797 vs. 0.722, p < 0.001) and sensitivity (71.4% vs. 52%) compared to the PET model.
Conclusion:
We developed a nomogram that noninvasively estimates the risk of LNM in lung cancer that may inform individualized preoperative assessment and evidence-based decision-making.
Key Points:
Question Can a nomogram integrating clinical, CT, and PET features improve preoperative prediction of lymph node metastasis in primary lung cancer, particularly in small nodes? Findings We developed a Clinical-CT-PET nomogram that achieved the best diagnostic accuracy (AUCs 0.883 and 0.877) among five models, especially for small lymph nodes (< 1 cm). Clinical relevance This noninvasive Clinical-CT-PET nomogram may improve the accuracy of preoperative lymph node staging and guide individualized treatment planning. It may also help avoid unnecessary invasive procedures in lung cancer patients, pending further multi-center validation.
