EVALUATION OF FDG-PET IMAGING FOR DIFFERENTIATING SEPTIC FROM ASEPTIC PAINFUL KNEE PROSTHESIS

Abstract

The accurate diagnosis of periprosthetic infection poses a challenge to the clinician and the imaging specialist alike. In recent years, FDG-PET imaging has shown great promise in the evaluation of occult infection at various anatomic sites. The purpose of this investigation was to determine the accuracy of FDG-PET imaging in diagnosing periprosthetic infection associated with total knee arthroplasty.

Sixty eight painful knee prostheses were referred for further evaluation with FDG-PET imaging. Approximately 60 minutes after the intravenous administration of FDG, PET images of both knees were acquired and interpreted by experienced nuclear medicine physicians. PET images demonstrating increased FDG activity at the bone-prosthesis interface were considered infected. Final diagnosis was made on the basis of surgical findings, histopathology, and clinical follow-up.

FDG-PET correctly diagnosed 19 of the 22 infected cases for a calculated sensitivity of 86.4% (19/22). FDG-PET correctly predicted the absence of infection in 38 of 46 aseptic knee prostheses for a calculated specificity of 82.6% (38/46). The negative and positive predictive values for FDG-PET imaging in this setting were 92.7% (38/41) and 70.4% (19/27), respectively. The overall accuracy of FDG-PET imaging was 83.8% (57/68). FDG-PET was indeterminate in three cases which were not included in this analysis.

These results demonstrate that FDG-PET is a useful diagnostic tool for the evaluation of possible infection associated with knee arthroplasty. Considering the large number of subjects who undergo total knee arthroplasty and the sizable fraction who develop complications following surgery, the impact of FDG-PET imaging could be substantial. Examination of a larger number of patients with painful knee prostheses will further clarify the merit of this powerful technique in this clinical setting.