Background/Objectives: To compare diagnostic performance and cost-effectiveness of 18FFDG PET/CT versus MRI for cervical lymph node assessment in nasopharyngeal carcinoma (NPC) and to evaluate their impact on N-staging and upper-neck-only irradiation planning. Materials and Methods: We retrospectively identified treatment-naïve NPC patients who underwent both MRI and FDG PET/CT within 14 days prior to ultrasound-guided biopsy of specific cervical lymph nodes with rigorous one-to-one multimodal matching. Using histopathology as the reference standard (Cohort A, node level), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the curve (AUC) were compared between PET/CT and MRI. In a staging cohort (Cohort B, patient level), we compared imaging-based N-staging accuracy and the pathology-concordant classification of treatment recommendations assuming upper-neck-only irradiation for N0 to N1 disease. In discordant cases (Cohort C), three experienced radiation oncologists designed dose prescriptions and neck irradiation volumes, first using MRI alone and then after reviewing PET/CT to quantify decision impact. A decision tree/Markov model (10-year horizon) evaluated cost-effectiveness of MRI- versus PET/CT-initiated strategies. Results: In total, 694 biopsy-verified lymph nodes from 649 patients were analyzed. PET/CT demonstrated higher sensitivity (96. 0% vs. 92. 6%, p = 0. 001) and NPV (80. 2% vs. 66. 7%, p < 0. 001) than MRI, with comparable specificity (64. 0% vs. 59. 0%, p = 0. 317) and PPV (91. 4% vs. 90. 0%, p = 0. 203) ; AUCs were 0. 864 and 0. 841, respectively (p = 0. 298). In Cohort B (N = 503), PET/CT provided accurate N-staging for a significantly higher proportion of patients compared to MRI (8. 0% vs. 4. 2%, p = 0. 021) and yielded more accurate recommendations for upper-neck-only irradiation when restricted to N0 to N1 disease (93. 8% vs. 88. 9%, p = 0. 003). In discordant cases (Cohort C, N = 62), PET/CT substantially improved accuracy compared with MRI and prompted clinically meaningful plan adjustments, including dose escalation for metastatic nodes (up to 16. 7%) and expansion from upper-neck-only to whole-neck irradiation with rates of 6. 4%, 8. 0%, and 11. 3% for the three radiation oncologists, respectively. In the base case economic analysis, PET/CT achieved higher effectiveness (5. 329 vs. 5. 305 quality-adjusted life years QALYs) at higher cost (US27, 228 vs. US25, 596), with an incremental cost–effectiveness ratio (ICER) of approximately US68, 000 per QALY, remaining below a willingness-to-pay threshold of US100, 000 per QALY; probabilistic sensitivity analysis favored PET/CT in 79. 6% of iterations. Conclusions: FDG PET/CT provided superior sensitivity and negative predictive value versus MRI for detecting nodal metastases in NPC, improving pathology-adjudicated N-staging and the accuracy of upper-neck-only irradiation recommendations. PET/CT was cost-effective in the modeled setting, although treatment de-escalation for benign nodes remained conservative in clinical decision-making.
Zhao et al. (Sat,) studied this question.