Impact of PET/computed tomography image reconstruction methods on Hopkins score in therapy response assessment of head and neck cancer patients.

PURPOSE: To evaluate the robustness of the Hopkins criteria score (HCS) with regards to new PET reconstruction methods in therapy response assessment of head and neck cancer after (chemo)radiotherapy [(C)RT].

METHODS

Maximum standardized uptake value and HCS were retrospectively assessed on fluorine-18-deoxyglucose PET/computed tomography (CT) 12 weeks after (C)RT for neck lymph nodes and the primary tumor site in 66 patients with three different reconstruction methods: standard high sensitivity reconstruction, European Association of Nuclear Medicine (EANM) Research Ltd. (EARL) reference standards 1 (EARL1) and 2 (EARL2). Histopathology (27 patients, 41%) or follow-up PET/CT 9 months after end of treatment (39 patients, 59%) served as reference standard. HCS 4 and 5 were rated as positive, 1-3 negative. The performance of HCS for all methods and intermethod concordance was assessed.

RESULTS

A total of 17 (26%) patients had tumor persistence. There was no significant difference in the performance of the reconstruction methods regarding tumor persistence. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy for PET/CT with HCS for overall tumor persistence was 77, 80, 57, 91, and 79% for sensitivity reconstruction, 65, 90, 69, 88, and 83% for EARL1 and 71, 82, 57, 89, and 79% for EARL2. The intermethod concordance of HCS across all sites demonstrated a substantial to perfect agreement with Cohens κ ranging from 0.6 to 1.0.

CONCLUSION

PET/CT HCS for therapy response assessment in head and neck cancer serves as a robust Score independent of newer sensitive reconstructions. Despite the high NPV of PET/CT in this situation the unsatisfactory PPV remains challenging.