18F]FDG PET/MRI versus contrast-enhanced MRI in detecting regional HNSCC metastases.

Abstract

To compare the accuracy of contrast-enhanced MRI using established dimensional and morphological criteria versus integrated [18F]FDG PET/MRI in identifying regional lymph node metastases in patients with newly diagnosed head and neck squamous cell carcinoma (HNSCC). For this purpose, we compare MRI and PET/MRI using the histopathological findings in dissected lymph nodes as the gold standard. We retrospectively reviewed 26 patients with histologically proven HNSCC who underwent gadolinium-enhanced [18F]FDG PET/MRI as part of their staging. All neck lymph nodes were classified on MRI using dimensional and/or morphological criteria. Then, they were jointly assessed by a nuclear medicine physician and a radiologist using integrated [18F]PET/MR images. ROC curves were obtained to compare the techniques. Lymph node histopathology was considered as the reference standard. Out of 865 lymph nodes, 35 were malignant at histopathology (3 with micro-metastases). Sensitivity and specificity were 48.6% and 99.5% for MRI using dimensional criteria; 60.0% and 99.6% for MRI using morphological criteria; 60.0% and 99.4% for MRI using both; and 74.3% and 97.6% for PET using MR as anatomic localization. The area under the ROC curve was higher for PET and MRI localization (0.859) than for MRI using dimensional (0.740; p < 0.05), or morphological (0.798; p < 0.05), or both criteria (0.797; p < 0.05). PET/MR using a PET SUVmax cutoff of 5.7 combined with MRI using dimensional and/or morphological criteria reached high values for accuracy (98.2%), NPV (98.2%), and PPV (95.2%). Compared with traditional contrast-enhanced MRI or PET alone, integrated PET/MRI could improve diagnostic accuracy in detecting metastatic lymph nodes in patients with HNSCC.