Is 18F-fluoride PET/CT an Accurate Tool to Diagnose Loosening After Total Joint Arthroplasty?

BACKGROUND Several studies using positron emission tomography (PET) show highly elevated periprosthetic bone uptake of fluorine-18 sodium fluoride (18F-fluoride), suggestive of implant loosening after arthroplasty. Focus so far has been on qualitative but not on quantitative assessment. There is also a lack of intraoperative confirmation of preoperative 18F-fluoride PET findings. Although the method seems to have acceptable accuracy and high sensitivity, an attempt to improve the specificity and an overall validation of the method appear warranted. QUESTIONS/PURPOSES (1) Is there a difference in 18F-fluoride uptake around loose versus well-fixed THA and TKA components? (2) Can 18F-fluoride uptake measures provide a threshold that differentiates loose from well-fixed implants undergoing revision for a variety of septic and aseptic indications? (3) In a population restricted to THA and TKA undergoing revision for aseptic indications, can measurement of 18F-fluoride uptake still distinguish loose from well-fixed components? (4) What is the interrater reliability of measuring 18F-fluoride uptake? METHODS This was a retrospective assessment of a diagnostic test, 18F-fluoride PET/CT, which was performed prior to revision surgery. We included 63 patients with 31 THAs and 32 TKAs. Sixty-five percent of patients were female, and the mean age at 18F-fluoride PET/CT was 66 years. The THA had different modes of fixation (cemented, cementless, and hybrid; 45%, 32%, and 23%, respectively), whereas all TKAs were cemented. Imaging was conducted using routine protocols 1 hour after tracer injection. The interobserver reproducibility was analyzed using Spearman rank correlations and Bland-Altman analyses. Two independent observers were trained separately by a nuclear physician to measure maximal periprosthetic standardized uptake values (SUVmax) for each arthroplasty component (n = 126). Findings at surgery (whether the components were well fixed or loose, as well as the presence or absence of infection) were used as a reference. Presence of periprosthetic joint infection was retrospectively determined based on the criteria suggested by the European Bone and Joint Infection Society (EBJIS): clinical features in combination with blood analysis, synovial fluid cytologic analysis, and microbiology test results. Receiver operating characteristic (ROC) curves were plotted to assess the area under the curve (AUC) for each investigated component separately, indicating suitable SUVmax thresholds that differentiate loose from well-fixed components. After excluding patients with confirmed or suspected PJI per the EBJIS criteria (n = 12), the above analysis was repeated for the remaining patients with aseptic loosening (n = 51). RESULTS We found higher 18F-fluoride uptake around loose versus well-fixed components in all but femoral TKA components (median [range] SUVmax for well-fixed versus loose THA cups 10 [7 to 30] versus 22 [6 to 64], difference of medians 12; p = 0.003; well-fixed versus loose TKA femoral components 14 [4 to 41] versus 19 [9 to 42], difference of medians 5; p = 0.38). We identified favorable ROC curves for all investigated components except femoral TKA components (THA cups AUC 0.81 [best threshold 13.9]; THA femoral stems AUC 0.9 [best threshold 17.3]; femoral TKA components AUC 0.6 [best threshold 14.3]; tibial TKA components AUC 0.83 [best threshold 15.8]). 18F-fluoride was even more accurate at diagnosing loosening when we limited the population to those patients believed not to have prosthetic joint infection (THA cups AUC 0.87 [best threshold 14.2]; THA femoral stems AUC 0.93 [best threshold 15.0]; femoral TKA components AUC 0.65 [best threshold 15.8]; tibial TKA components AUC 0.86 [best threshold 14.7]). We found strong interrater correlation when assessing SUVmax values, with Spearman ρ values ranging from 0.96 to 0.99 and Bland-Altman plots indicating excellent agreement between the two independent observers. CONCLUSION Measuring SUVmax after 18F-fluoride PET/CT is a useful adjunct in the diagnostic evaluation for suspected implant loosening after THA and TKA. The method appears to be both accurate and reliable in diagnosing implant loosening for all components except femoral TKA components. In a real-world mixed population with both low-grade infection and aseptic loosening, the method seems to be fairly easy to learn and helpful to subspecialized arthroplasty clinicians. When infection can be ruled out, the method probably performs even better. Further prospective studies are warranted to explore the reason why femoral TKA component loosening was more difficult to ascertain using this novel technique. LEVEL OF EVIDENCE Level III, diagnostic study.