Fever of Unknown Origin (FUO / PUO)
FDG PET/CT is a powerful whole-body case-finder in fever (or inflammation) of unknown origin, pointing to the cause in roughly half of patients. Reported diagnostic yield is about 50–60% (decisive in ≈54% in a large series), with overall sensitivity ≈80% and specificity ≈90%. Because FDG accumulates in infection, inflammation and malignancy alike, it is a highly sensitive “where to look next” test rather than a specific diagnosis.
Classic FUO is prolonged fever without a diagnosis after appropriate initial work-up. FDG PET/CT surveys the whole body for a metabolic focus — vasculitis, occult abscess, endocarditis, lymphoma or solid tumour — directing biopsy or further imaging. A normal scan has a high negative predictive value and supports a benign, self-limiting course.
Clinical context & when to image
- Classic fever of unknown origin after an appropriate first-line work-up.
- Inflammation of unknown origin (raised markers without localising features).
- Searching for a focus — occult infection, vasculitis or malignancy.
- Directing biopsy or targeted imaging to the most metabolically active site.
Imaging strategy
- FDG PET/CT is the preferred single whole-body test after the initial work-up.
- Integrate with history, examination, cultures and serology.
- A normal study has high negative predictive value and is reassuring.
- Follow abnormal foci with targeted confirmation.
Protocol
- Standard whole-body FDG PET/CT with glycaemic control, ideally before or within 3 days of starting glucocorticoids; add myocardial suppression if a cardiac cause is possible.
- Head-to-toe coverage where vasculitis or peripheral foci are possible.
- Correlate with any recent cross-sectional imaging.
How to read it
- Large-vessel wall uptake → vasculitis; nodal/organ uptake → lymphoma or tumour; focal collection → abscess; valve/device uptake → endocarditis.
- Interpret each focus in clinical context.
- Pursue the most active/accessible site for tissue or culture.
Diagnostic performance
In a large retrospective single-centre series (300 patients) FDG PET/CT was decisive in ≈54% of FUO/IUO cases, with overall sensitivity 80.2% and specificity 89.8%. Performance by cause is shown below; systematic reviews report an overall diagnostic yield of roughly 50–60%.
| Cause found | Sensitivity | Specificity |
|---|---|---|
| Overall (any cause) | ≈80% | ≈90% |
| Infection | ≈85% | ≈99% |
| Malignancy | ≈73% | ≈100% |
| Autoimmune / rheumatic | ≈85% | ≈99% |
Pitfalls
- Non-specific — physiological and benign inflammatory uptake are common.
- Recent infection, granulomas and brown fat cause false positives.
- Findings require directed confirmation (biopsy/culture).
- A negative scan does not exclude every cause but is reassuring.
Evidence & guidelines
- Meta-analyses report a pooled diagnostic yield of 56–58% in FUO.
- Large series: decisive in ≈54%; overall sensitivity ≈80%, specificity ≈90%.
- Expert reviews support FDG PET/CT as a high-yield step after the initial work-up.
In depth
- FDG PET/CT is the nuclear medicine test of choice for FUO; a meta-analysis of 23 studies (1927 patients with FUO or inflammation of unknown origin) found pooled sensitivity of 84% and specificity of 63%.
- Classic FUO is fever of 38.3 °C or higher on several occasions for at least 3 weeks without a diagnosis despite appropriate investigation; raised inflammatory markers with temperature not exceeding 38.3 °C is termed inflammation of unknown origin (IUO).
- FUO has many causes in four main groups (infection, malignancy, non-infectious inflammatory disease and miscellaneous), and about half of cases remain undiagnosed; mortality has fallen from 21–33% before 1980 to about 7% in recent series.
- ⁶⁷Ga-citrate and labelled leukocyte scintigraphy have lower diagnostic performance than FDG PET/CT and are now used mainly when PET/CT is unavailable; leukocyte imaging does not detect most malignancies or non-infectious inflammation.
- WBC scintigraphy is best reserved for cases with clues suggesting focal infection.
- In children with FUO, FDG PET/CT performs similarly to adults, with pooled sensitivity of about 83% and specificity of about 78%; an abnormal scan makes a definite diagnosis much more likely.
Sources: Kan et al. Acta Radiol 2019 (PMID 30205705) · Palestro et al. AJR 2023 (PMID 36722759) · Petersdorf & Beeson 1961 and Durack & Street 1991 criteria, as summarised by Minamimoto, Jpn J Radiol 2022 (PMID 35781177) · Minamimoto, Jpn J Radiol 2022 (PMID 35781177) · Martinello et al. umbrella review, Q J Nucl Med Mol Imaging 2026 (PMID 41873478) · Li et al. Pediatr Radiol 2022 (PMID 35348809)
- Fever and weight loss in a 71-year-old woman · Intermediate