Evaluation of neck and body metastases to nodes with ferumoxtran 10-enhanced MR imaging: phase III safety and efficacy study.

PURPOSE: To determine the safety and efficacy of ferumoxtran 10-enhanced magnetic resonance (MR) imaging for diagnosis of metastases to lymph nodes and the clinical usefulness of ferumoxtran 10 in nodal staging. MATERIALS AND METHODS: One hundred fifty-two patients were injected with ferumoxtran 10. Readers independently evaluated precontrast MR images by using node size criteria and subjective assessment of other imaging features. Ferumoxtran 10-enhanced MR images were evaluated alone and paired with precontrast images for comparison. The diagnostic performances of precontrast MR size criteria and postcontrast MR imaging were evaluated with receiver operating characteristic (ROC) analysis. Lymph node signal intensity was correlated with histopathologic findings. MR imaging and histopathologic nodal stages were compared. RESULTS: Node-level sensitivity, specificity, and accuracy of precontrast MR imaging were 54%, 82%, and 68%, respectively, with node size criterion alone; 91%, 51%, and 71%, respectively, with subjective reader assessment; 85%, 85%, and 85%, respectively, with postcontrast MR imaging alone; and 83%, 77%, and 80%, respectively, with paired pre- and postcontrast MR imaging. Compared with size criteria, subjective reader assessment had higher sensitivity but substantially lower specificity. Areas under the ROC curve for pre- and postcontrast MR imaging were 0.76 and 0.83, respectively. Nonmetastatic nodes had significantly lower signal intensity than metastatic nodes on postcontrast T2-weighted MR images (P <.001). Postcontrast nodal staging was significantly more accurate than precontrast nodal staging (P <.01). Headache, back pain, vasodilatation, and urticaria each occurred in 6% of patients. CONCLUSION: Ferumoxtran 10-enhanced MR imaging was safe and effective and facilitated improved diagnostic performance. Use of iron oxide-enhanced MR imaging increased the positive predictive value by 20% and the accuracy by 14% compared with reader assessment. Differentiating patients with no nodal metastatic involvement was more reliable with ferumoxtran 10-enhanced MR imaging than with precontrast MR imaging.