Related Experiment Video
Updated: Jun 20, 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
Lymphotropic nanoparticle-enhanced MRI for independent prediction of lymph node malignancy: a logistic regression
Pari V Pandharipande1, Jose T Mora, Raul N Uppot
1Department of Radiology, Institute for Technology Assessment, Massachusetts General Hospital, 101 Merrimac St., 10th fl., Boston, MA 02114, USA.
Objective:
The purpose of this study was to determine whether use of lymphotropic nanoparticle-enhanced MRI can improve the ability to characterize lymph nodes as benign or malignant beyond size criteria alone.
Materials And Methods:
The cases of 42 consecutively registered patients with a known primary malignant tumor of the genitourinary tract who underwent both lymphotropic nanoparticle-enhanced MRI and CT-guided biopsy of a lymph node at our institution from 2000 to 2005 were retrospectively identified. Lymphotropic nanoparticle-enhanced MRI included T2(*)-weighted gradient-recalled echo imaging before and 24-36 hours after i.v. administration of lymphotropic iron oxide nanoparticles. Two positivity criteria for lymph node malignancy were evaluated independently: lack of nanoparticle uptake at lymphotropic nanoparticle-enhanced MRI and short-axis length of 1 cm or greater. Sensitivity and specificity were calculated for each criterion with biopsy results as the standard of reference. Logistic regression analysis was used to determine the association (odds ratio) between lymphotropic nanoparticle-enhanced MRI findings and the presence of lymph node malignancy when controlling for short-axis length.
Results:
Metastatic lesions were detected at histologic examination in 67% (28/42) of nodes. According to the lymphotropic nanoparticle-enhanced MRI criterion, sensitivity for malignancy was 100% (28/28 nodes), and specificity was 64% (9/14 nodes). According to the short-axis criterion, sensitivity was 79% (22/28 nodes), and specificity was 21% (3/14 nodes). In multivariate analysis, when controlling for short-axis length, the finding of malignancy at lymphotropic nanoparticle-enhanced MRI was an independent predictor of the presence of malignancy (odds ratio, 61.0; 95% CI, 8.0 to infinity; p < 0.0001).
Conclusion:
Use of lymphotropic nanoparticle-enhanced MRI may improve ability to characterize lymph nodes beyond size criteria alone. Our results emphasize the need to further assess lymphotropic nanoparticle-enhanced MRI in prospective large-scale studies with wider variation in the distribution of lymph node sizes and primary malignancies.
