• BACKGROUND
    • Radiographs are the standard clinical tool used to monitor patients with pelvic osteolysis after total hip arthroplasty; however, previous reports have questioned the value and accuracy of this method. With use of a cadaveric model, we investigated the accuracy of radiographs in determining the location and size of periacetabular osteolysis.
  • METHODS
    • We implanted total hip arthroplasty components in eight cadaveric hips and made four radiographs of each hip from different views. We then removed the components and created two pelvic defects in each hip. We measured the volume of each defect, reimplanted the components, and made another set of radiographs. The defects were then enlarged two more times, with the volume measured and another set of radiographs made each time. In total, 128 radiographs were made of forty-eight lesions. An orthopaedist who was blinded to the location of the lesions assessed the radiographs with regard to the presence and size of osteolytic lesions.
  • RESULTS
    • The overall sensitivity for the detection of osteolysis on a single radiograph was 41.5%, and the overall specificity was 93.0%. Sensitivity was dependent on the location and size of the lesions but not on the radiographic view. Sensitivity ranged from 72% for lesions in the ilium to 10 cm (3) was significantly higher than that for smaller lesions (p < 0.001). When all four radiographic views of one lesion were analyzed together, sensitivity increased to 73.6%. Despite the low sensitivity, specificity remained high, indicating that once osteolysis is evident radiographically, the likelihood that a lesion truly exists is high. Additionally, we found that the extent of osteolysis was substantially underestimated on radiographs.
  • CONCLUSIONS
    • The use of radiographs to assess and monitor osteolysis has both limitations and merits. Using multiple views, an experienced orthopaedist identified only 73.6% of pelvic lesions. However, once a pelvic osteolytic lesion is evident radiographically, the likelihood that it truly exists is high.