Introduction: This prospective study was designed to elucidate the advantages and pitfalls of (18)F-FDG PET in detecting locally residual/recurrent nasopharyngeal carcinoma (NPC) in comparison with MRI.
Methods: We recruited NPC patients from two ongoing prospective trials. One is being performed to evaluate suspected local recurrence (group A) and the other to assess local treatment response 3 months after therapy (group B). Both groups received (18)F-FDG PET and head and neck MRI. The gold standard was histopathology or clinical/imaging follow-up. An optimal cut-off standardised uptake value (SUV) was retrospectively determined.
Results: From January 2002 to August 2004, 146 patients were eligible. Thirty-four were from group A and 112 from group B. In all, 26 had locally recurrent/residual tumours. Differences in detection rate between (18)F-FDG PET and MRI were not statistically significant in either group. However, (18)F-FDG PET showed significantly higher specificity than MRI in detecting residual tumours among patients with initial T4 disease (p=0.04). In contrast, the specificity of (18)F-FDG PET for patients with an initial T1-2 tumour treated with intracavitary brachytherapy (ICBT) was significantly lower than that for patients not treated by ICBT (72.2% vs 98.1%, p=0.003). At an SUV cut-off of 4.2, PET showed an equal and a higher accuracy compared with MRI in groups A and B, respectively.
Conclusion: (18)F-FDG PET is superior to MRI in identifying locally residual NPC among patients with initial T4 disease but demonstrates limitations in assessing treatment response in patients with initial T1-2 disease after ICBT. A cut-off SUV is a useful index for aiding in the visual detection of locally residual/recurrent NPC.