Local relapse of nasopharyngeal cancer and Voxel-based analysis of FMISO uptake using PET with semiconductor detectors

Background: Hypoxic cancer cells are thought to be radioresistant and could impact local recurrence after radiotherapy (RT). One of the major hypoxic imaging modalities is [¹⁸F]fluoromisonidazole positron emission tomography (FMISO-PET). High FMISO uptake before RT could indicate radioresistant sites and might be associated with future local recurrence. The predictive value of FMISO-PET for intra-tumoral recurrence regions was evaluated using high-resolution semiconductor detectors in patients with nasopharyngeal carcinoma after intensity-modulated radiotherapy (IMRT). Methods: Nine patients with local recurrence and 12 patients without local recurrence for more than 3years were included in this study. These patients received homogeneous and standard doses of radiation to the primary tumor irrespective of FMISO uptake. The FMISO-PET image before RT was examined via a voxel-based analysis, which focused on the relationship between the degree of FMISO uptake and recurrence region. Results: In the pretreatment FMISO-PET images, the tumor-to-muscle ratio (TMR) of FMISO in the voxels of the tumor recurrence region was significantly higher than that of the non-recurrence region (p<0.0001). In the recurrent patient group, a TMR value of 1.37 (95% CI: 1.36-1.39) corresponded to a recurrence rate of 30%, the odds ratio was 5.18 (4.87-5.51), and the area under the curve (AUC) of the receiver operating characteristic curve was 0.613. In all 21 patients, a TMR value of 2.42 (2.36-2.49) corresponded to an estimated recurrence rate of 30%, and the AUC was only 0.591. Conclusions: The uptake of FMISO in the recurrent region was significantly higher than that in the non-recurrent region. However, the predictive value of FMISO-PET before IMRT is not sufficient for up-front dose escalation for the intra-tumoral high-uptake region of FMISO. Because of the higher mean TMR of the recurrence region, a new hypoxic imaging method is needed to improve the sensitivity and specificity for hypoxia.