Objectives: To describe etiologies of classic fever of unknown origin (FUO) and evaluate associations of FDG PET/CT, C-reactive protein (CRP), age, and Charlson Comorbidity Index (CCI) with biopsy performance, diagnostic yield, and time to diagnosis.
Methods: Retrospective cohort of 179 adults with classic FUO. In the PET/CT subgroup (n = 106), logistic regression assessed predictors of biopsy and final diagnosis; CRP discrimination was evaluated by ROC analysis; and time to diagnosis by Kaplan-Meier analysis (CCI=0 vs CCI≥3).
Results: Autoimmune/autoinflammatory diseases (31.3%), infections (30.7%), and malignancy (19.0%) predominated; 16.8% remained undiagnosed. PET/CT positivity predicted biopsy (aOR 4.85, 95% CI: 1.45-16.19) but not diagnostic yield. Higher log-CRP increased odds of diagnosis (OR 2.11, 95% CI: 1.28-3.48), whereas age decreased odds (OR 0.95 per year, 95% CI: 0.90-0.99). CRP AUC 0.68 (cut-off 30.3 mg/L). CCI≥3 was associated with longer time to diagnosis (log-rank P = 0.034).
Conclusion: PET/CT mainly facilitated biopsy, while diagnostic yield was more closely related to CRP and age. High comorbidity burden may contribute to diagnostic delay.
Keywords: 18F-FDG PET/CT; C-reactive protein; Charlson Comorbidity Index; Diagnostic delay; Fever of unknown origin.
Copyright © 2026 The Authors. Published by Elsevier Ltd.. All rights reserved.