Detection of infected hip arthroplasty using FDG-PET

Abstract

1152

Objectives: Clinical determination of prosthetic joint infection remains challenging. FDG-PET has shown promise to detect infection. This serves to update our prior paper with new results using FDG-PET to non-invasively identify infection in painful hip prostheses.

Methods: FDG-PET of the affected hip(s) was performed in 98 patients with painful prostheses when clinical and laboratory data failed to determine an infectious cause. Images were interpreted in blinded fashion, with infection identified by increased FDG uptake at the bone-prosthesis interface versus adjacent tissue. A second blinded read by a single senior nuclear medicine radiologist was performed to eliminate interobserver variability. Clinical outcome was available for 82 patients (84 hips) of 98 enrolled, determined by surgical, laboratory, and clinical follow-up data.

Results: Of 84 hip prostheses with full follow-up, FDG-PET identified prosthetic infection in 13 of 16 (sens=81.3%), with false positive (FP) results in 17 of 68 (spec=75.0%). In the second blinded read, sensitivity improved to 93.8% (15/16), and specificity to 79.4% (FP in 14/68). Six of 14 FP reads showed extensive osteolysis or periprosthetic fracture non-union visible on plain film radiography; when these were removed, specificity improved to 87.1% (54/62). These final FDG-PET data, when patients with a mechanical cause for pain were removed, had a PPV of 65.2% (15/23), NPV of 98.2 % (54/55), and accuracy of 88.5% (69/78).

Conclusions: When interpreted by experienced radiologists, FDG-PET is very sensitive and specific for assessment of infection in painful hip arthroplasty. Specificity is decreased in patients with radiologic evidence of mechanical hip pain, and must be interpreted with caution.

Research Support: Grant from NIH