Dynamic Whole-Body FDG PET/CT for Predicting Malignancy in Head and Neck Tumors and Cervical Lymphadenopathy.

Background: Dynamic whole-body (D-WB) FDG PET/CT is a novel technique that enables the direct reconstruction of multiparametric images representing the FDG metabolic uptake rate (MRFDG) and "free" FDG (DVFDG). Applying complementary parameters with distinct characteristics compared to static SUV images, the aims of this study are as follows: (1) to determine the threshold values of SUV, MRFDG, and DVFDG for malignant and benign lesions; (2) to compare the specificity of MRFDG and DVFDG images with static SUVbw images; and (3) to assess whether any of the dynamic imaging parameters correlate more significantly with malignancy or non-malignancy in the examined lesions based on the measured values obtained from D-WB FDG PET/CT. Methods: The study was a retrospective analysis of D-WB PET/CT data from 43 patients (23 males and 20 females) included both in the context of primary staging as well as imaging performed due to suspicion of post-therapeutic relapse or recurrence. Standard scanning was performed using a multiparametric PET acquisition protocol on a Siemens Biograph Vision 600 PET/CT scanner. Pathological findings were manually delineated, and values for SUVbw, MRFDG, and DVFDG were extracted. The findings were classified and statistically evaluated based on their was histological verification of a malignant or benign lesion. Multinomial and binomial logistic regression analyses were used to find parameters for data classification in different models, employing various combinations of the input data (SUVbw, MRFDG, DVFDG). ROC curves were generated by changing the threshold p-value in the regression models to compare the models and determine the optimal thresholds. Results: Patlak PET parameters (MRFDG and DVFDG) combined with mean SUVbw achieved the highest diagnostic accuracy of 0.82 (95% CI 0.75-0.89) for malignancy detection (F1-score = 0.90). Sensitivity reached 0.85 (95% CI 0.77-0.91) and specificity 0.93 (95% CI 0.87-0.98). Classification accuracy in tumors was 0.86 (95% CI 0.78-0.92) and in lymph nodes 0.81 (95% CI 0.73-0.88). Relative contribution analysis showed that DVFDG accounted for up to 65% of the classification weight. ROC analysis demonstrated AUC values above 0.8 for all models, with optimal thresholds achieving sensitivities of around 0.85 and specificities up to 0.93. Thresholds for malignancy detection were, for mean values, SUVbw > 5.8 g/mL, MRFDG > 0.05 µmol/mL/min, DVFDG > 68%, and, for maximal values, SUVbw > 8.7 g/mL, MRFDG > 0.11 µmol/mL/min, DVFDG > 202%. Conclusions: The D-WB [18F]FDG PET/CT images in this study highlight the potential for improved differentiation between malignant and benign lesions compared to conventional SUVbw imaging in patients with locally advanced head and neck cancers presenting with cervical lymphadenopathy and carcinoma of unknown primary origin (CUP). This observation may be particularly relevant in common diagnostic dilemmas, especially in distinguishing residual or recurrent tumors from post-radiotherapy changes. Further validation in larger cohorts with histopathological confirmation is warranted, as the small sample size in this study may limit the generalizability of the findings.