Prosthetic Joint Infection
Suspected infection of a hip or knee prosthesis demands a specific test, because peri-prosthetic marrow and post-operative remodelling mimic infection. Labelled-leucocyte imaging, with a bone-marrow scan in equivocal cases, is the most accurate nuclear test — combined WBC/marrow accuracy exceeds 90% in most series (sensitivity 100%, specificity 91%, accuracy 95% in one comparative study) — interpreted on the WBC/marrow mismatch principle. FDG PET/CT is highly sensitive (absence of peri-prosthetic uptake reliably excludes infection) but less specific than WBC imaging; the EANM/SNMMI guideline supports it mainly for hip prostheses.
A three-phase bone scan is sensitive but cannot separate infection from aseptic loosening, so it mainly serves to exclude other causes of a painful prosthesis. The decisive test compares the distribution of labelled white cells against a bone-marrow map: infection produces white-cell activity that does not correspond to marrow (incongruence), whereas displaced or expanded marrow produces matching (congruent) uptake.
Clinical context & when to image
- A painful prosthesis with clinical or biochemical suspicion (raised CRP/ESR) and equivocal radiographs.
- Differentiating septic from aseptic loosening before revision surgery.
- When joint aspiration is inconclusive or dry.
- As part of a multidisciplinary work-up alongside aspiration, microbiology and inflammatory markers.
Imaging strategy
- Radiographs first; a negative three-phase bone scan excludes PJI, but peri-prosthetic remodelling makes a positive scan non-specific, especially in the first years after implantation.
- Labelled-leucocyte imaging (⁹⁹ᵐTc-HMPAO or ¹¹¹In-oxine), combined with ⁹⁹ᵐTc-colloid (nanocolloid or sulfur colloid) marrow imaging when the WBC study is equivocal, is the most accurate nuclear test.
- Antigranulocyte-antibody scintigraphy is an alternative where in-vitro cell labelling is unavailable.
- FDG PET/CT is sensitive and convenient but specificity varies with interpretation criteria; EANM/SNMMI v2.0 supports it for suspected hip PJI (preferably ≥2 years after surgery) and for ruling out infection when pre-test probability is low, with insufficient evidence for knees and shoulders.
- SPECT/CT improves anatomical localisation.
Protocol
- Autologous leucocytes labelled with ⁹⁹ᵐTc-HMPAO (or ¹¹¹In-oxine) and re-injected; EANM protocol: early (30 min–1 h), delayed (2–4 h) and late (20–24 h) images, acquired decay-corrected and displayed on the same absolute scale so change over time can be judged.
- ⁹⁹ᵐTc-colloid marrow scan (EANM: in equivocal WBC studies), acquired in the same views for direct spatial comparison with the white-cell study.
- SPECT/CT of the prosthesis for localisation.
- Interpret white-cell and marrow images side by side in the same planes.
How to read it
- Congruent (white-cell uptake spatially matches marrow) → displaced/expanded marrow, not infection.
- Incongruent (white-cell uptake without corresponding marrow) → infection.
- Increasing intensity or extent from delayed (3–4 h) to late (20–24 h) white-cell images favours infection; decreasing activity argues against it.
- Correlate with SPECT/CT to localise activity to the bone–prosthesis interface versus soft tissue.
Diagnostic performance
In a comparison in 59 failed lower-limb joint replacements, combined ¹¹¹In-leucocyte / ⁹⁹ᵐTc-sulfur-colloid marrow imaging achieved sensitivity 100%, specificity 91% and accuracy 95%, out-performing coincidence-detection FDG imaging (best accuracy 71% across four interpretation criteria). A 2022 meta-analysis of FDG PET/CT (23 studies) reported pooled sensitivity ≈85% and specificity ≈86%, with wide heterogeneity between studies.
| Modality | Sensitivity | Specificity | Role |
|---|---|---|---|
| Combined WBC / marrow | ≈100% | ≈91% (accuracy ≈95%) | Most accurate nuclear test — imaging modality of choice |
| FDG PET/CT | ≈85% (pooled) | ≈86% (pooled; heterogeneous) | High NPV; mainly hip prostheses |
| Three-phase bone scan | High | Low | Sensitive screen; a negative scan excludes PJI |
| Antigranulocyte antibody | ≈83% | ≈80% | Alternative to in-vitro cell labelling |
Pitfalls
- Recent surgery alters marrow distribution and causes peri-prosthetic remodelling (bone scan) and reactive FDG uptake — record the date of surgery; labelled leucocytes do not usually accumulate in normally healing wounds.
- Marrow packing or expansion around prostheses causes matched uptake — the reason the marrow scan is essential.
- FDG uptake can persist around normal prostheses and in aseptic loosening, lowering specificity.
- Single-time-point or non-decay-corrected WBC acquisitions, and equivocal studies read without a marrow scan, cause false results.
Evidence & guidelines
- Combined leucocyte/marrow imaging is the most accurate radionuclide test for PJI (prospective and comparative data).
- The EANM/EBJIS/ESR consensus (2019) proposes WBC scintigraphy (± marrow scan), anti-granulocyte scintigraphy or non-contrast MRI as the initial advanced imaging test within a multidisciplinary pathway, noting that FDG PET has high sensitivity but lower specificity than WBC imaging.
- The EANM/SNMMI FDG guideline (v2.0, 2024) lists suspected hip PJI and ruling out PJI in low-probability cases as indications; evidence for knees, shoulders and ankles is insufficient.
In depth
- Combined WBC/marrow imaging relies on spatial congruence: labelled leucocytes and colloid both accumulate in marrow, but infection stimulates leucocyte uptake while suppressing colloid uptake, so an incongruent pattern (WBC activity without matching marrow activity) indicates infection.
- Pooled WBC/marrow performance: hip 69% sensitivity and 96% specificity; knee 80% sensitivity and 93% specificity.
- Periprosthetic infection timing follows 'thirds': about one third early (<3 months) and one third delayed (3–12 months), mostly from organisms introduced at surgery, and the rest late (>1 year), usually haematogenous.
- The bone scan is best used as a screen, because a negative scan makes infection and aseptic loosening unlikely, but its specificity is only about 69% at the hip and 56% at the knee.
- FDG PET meta-analyses report sensitivity of about 70–86% and specificity of about 84–93% for lower-limb periprosthetic infection (hip 86%/93%, knee 70%/84%), and there are still no standardised interpretive criteria.
- In a small prospective study (34 hip/knee prostheses), ⁶⁸Ga-citrate PET/CT was more specific than FDG PET/CT (88% vs 38%) for separating infection from sterile inflammation, with similar sensitivity; data remain preliminary.
- In shoulder arthroplasty, where chronic low-grade infection (often Cutibacterium) predominates, WBC/marrow SPECT/CT and FDG PET/CT were specific but insensitive (sensitivity 18% and 14%), so they act only as 'rule-in' tests.
Sources: PMID 36496268 (Palestro, Semin Nucl Med 2023) · PMID 27707850 (Verberne, JBJS Am 2016) · PMID 28050818 (Verberne, Clin Orthop Relat Res 2017) · PMID 36496268 (Palestro 2023) · PMID 15483283 (Zimmerli, NEJM 2004) · PMID 27707850 · PMID 28050818 · PMID 36496268
- A painful hip replacement · Intermediate