There are limited data on evaluating the significance of radiolucent lines and aseptic loosening in total knee arthroplasty (TKA). We sought to compare the sensitivity, specificity, and reliability of the Knee Society Total Knee Arthroplasty Roentgenographic Evaluation and Scoring System (KSRES) in detecting tibial component loosening compared to a novel percentage-based system (PBS).
We retrospectively reviewed radiographs obtained from 48 patients within 6 months prior to revision TKA. The radiographs were randomized and four reviewers independently used the KSRES to categorize tibial implants as nonconcerning, clinical follow-up for progression, or loose as described by KSRES. For the PBS, the percent involvement of the tibial implant interface of any radiolucency at the bone-cement or cement-implant interface was determined. The higher percentage from either the anteroposterior or the lateral image was the final score. Components were categorized as nonconcerning (≤10%), clinical follow-up for progression (11%-24%), or loose (≥25%). We compared the sensitivity, specificity, and interobserver reliability using intraoperative assessment of implant fixation as the gold standard.
For the KSRES, the mean sensitivity for determining tibial loosening was 7.3% and mean specificity for determining a nonconcerning implant was 95.9%. The PBS significantly increased the sensitivity to 91.1% (P < .001) while maintaining a specificity of 87.9% (P = .2). Interobserver reliability significantly increased from a mean kappa of 0.26 to 0.75 (P < .001).
The KSRES significantly underestimates implant loosening. The proposed percentage-based system demonstrated excellent sensitivity, specificity, and interobserver reliability in determining tibial implant loosening in this patient population.