Discrepancy between X-ray and CT assessment in the Mirels score for long bone metastasis

The Mirels score is used widely to predict pathological fractures in long bone metastases. Four factors are used in this scoring system, assessed based on patient complaints and radiographic images: pain, (anatomical) site, size, and lesion (radiographic appearance). However, in recent clinical practice, orthopaedic surgeons are required to evaluate the fracture risk of long bone metastases solely using CT images. The consistency of the Mirels scores between X-ray and CT assessments is understudied. This study aimed to examine the agreement between radiographic and CT assessments for three factors (site, size, and lesion). We retrospectively analyzed the Mireles scores of long bone metastases with radiographic and CT images taken during the same period between 2013 and 2024. Anatomical site, size, and radiographic appearance were independently evaluated through a blinded test by three observers (an orthopaedic oncologist, a board-certified general orthopaedic surgeon, and an orthopaedic senior resident). The kappa coefficient of agreement was evaluated using the Landis criterion. Fifty-two cases (22 females and 30 males) with a mean age of 67 years (range 48–86) were analyzed, including 39 femurs, seven tibias, and six humeri. Kappa values for the intra-observer variability of the Mirels score between X-ray and CT were moderate to good(0.41–0.67)for lesion but poor to fair (0.17–0.34) for size. CT assessments tended to underestimate the size compared with X-rays across all observers; the average size score was 2.35 (SD 0.78) for X-ray versus 2.02 (SD 0.87) for CT (p < 0.0001). The Mirels score assessments were consistent between X-ray and CT for lesion but not for size, and CT tended to underestimate the size compared to X-ray assessment. Therefore, clinicians should be cautious of using only CT-based axial measurements in predicting the risk of fracture.