Effectiveness of head-and-neck molecular imaging reporting and data system criterion in head-and-neck squamous cell carcinoma postconcurrent chemoradiotherapy

Purpose: Postconcurrent chemoradiotherapy (CRT) response assessment has been challenging in locally advanced head-and-neck squamous cell carcinoma (LA-HNSCC) due to prevailing postradiation changes. Molecular response methods have been encouraging, although further clarifications and validations were needed. We tested the effectiveness of a proposed semi-quantitative molecular response criterion in these patients. Materials and Methods: Two subspecialty-trained physicians evaluated F-18-fluorodeoxyglucose positron emission tomography/computed tomography of LA-HNSCC patients (n = 83) post 3 months CRT using a five points Head and Neck Molecular Imaging-Reporting and Data System (HAN-MI-RADS) criterion. Where available, histopathology examination with clinical and imaging interpretation was taken as a reference for the disease. A diagnostic accuracy comparison was done with the existing Hopkins score. Further effectiveness was analyzed with disease-free survival (DFI) and overall survival (OS). Results: Metastasis was developed in 11/83 patients at 3 months of evaluation. Of 72 patients, 39, 2, and 31 patients had a complete response, equivocal response, and partial response as per HAN-MI-RADS. Per patient sensitivity, specificity, positive predictive value, negative predictive value, and accuracy for predicting loco-regional disease up to 1 and 2 years was 93.3%, 92.5%, 90.3%, 94.9%, 92.9%, and 84.9%, 91.9%, 90.3%, 87.2%, and 88.6% respectively. One year and two years DFI for each HAN-MI-RADS score showed a statistically significant difference while it was not for OS. The receiver operating characteristic curve analysis showed significantly better outcome predictability of HAN-MI-RADS (area under the curve [AUC] 0.884) than Hopkins (AUC 0.699). Conclusions: A five points HAN-MI-RADS criterion was found promising for response assessment with less equivocal results and statistically significant higher AUC than Hopkins for loco-regional recurrence prediction.