• Oliver Seitz, Nicole Chambron-Pinho, Markus Middendorp, Rober Sader, Martin Mack, Thomas J Vogl, and Sotirios Bisdas.
• Department of Oromaxillofacial Surgery, JW Goethe University, Frankfurt, Germany.
• Neuroradiology. 2009 Oct 1; 51 (10): 677-86.

IntroductionThe purpose of this paper is to evaluate the impact of adding combined 18F-PET/CT to MRI for T and N staging of the oral and oropharyngeal cancer and calculation of the gross tumor volume (GTV) having histopathology as reference standard.MethodsPET/CT and MRI were performed in 66 patients with suspected oral and oropharyngeal cancer (41 primary tumors/25 recurrent tumors) and nodal disease (114 nodes). Statistical analysis included the McNemar test, sensitivity, specificity for the diagnostic modalities as well as regression analysis, and Bland-Altman graphs for calculated tumor volumes.ResultsThere was no statistically significant difference between the two modalities compared to pathological findings regarding detection of disease (P > or =0.72). The sensitivity/specificity for tumor detection were 100/80% and 96.72/60% for MRI and PET/CT, respectively. The sensitivity/specificity for nodal metastases were 88.46/75% and 83.81/73.91% for MRI and PET/CT, respectively. In 18% of cases, the MRI-based T staging resulted in an overestimation of the pathologic tumor stage. The corresponding rate for PET/CT was 22%. Regarding the treated necks, both modalities showed 100% sensitivity for detection of the recurrent lesions. In necks with histologically N0 staging, MRI and PET/CT gave 22% and 26% false positive findings, respectively. The mean tumor volume in the pathologic specimen was 16.6 +/- 18.6 ml, the mean volume derived by the MR imaging was 17.6 +/- 19.1 ml while the estimated by PET/CT volume was 18.8 +/- 18.1 ml (P < or =0.007 between the three methods). The Bland-Altman analysis showed a better agreement between PET/CT and MRI.ConclusionThe diagnostic performance of FDG-PET/CT in the local staging of oral cancer is not superior to MRI.

36 keywords...

hide…