Bone-RADS caught 97% of malignant lytic lesions, at the cost of specificity

Design

Retrospective external validation, single centre, four independent readers

Population

423 lytic long-bone lesions in skeletally mature patients, mean age 62

Primary outcome

Malignancy on biopsy or stability over two years

Effect

Sensitivity 97.1%, specificity 67.4%, PPV 85.2%, NPV 92.4%; OR 9.7 per score point

A retrospective external validation in the Journal of Bone and Joint Surgery, published on 30 September, applied the American College of Radiology's Bone-RADS score to 423 lytic lesions in the long bones of adults (mean age 62) found on 2021 radiographs. Four reviewers scored each lesion independently, and lesions were counted benign if biopsy showed so or if they were stable for two years.

Each one-point rise in Bone-RADS score raised the odds of malignancy about tenfold (OR 9.7, 95% CI 6.2 to 14.9). For separating possibly benign from malignant, sensitivity was 97.1% and specificity 67.4%, with a positive predictive value of 85.2% and negative predictive value of 92.4%. Lesion margin, endosteal erosion, pathological fracture and a soft-tissue mass each carried risk. Agreement between readers on the full four-tier score was only moderate (weighted kappa 0.52), but better for the simple benign-versus-malignant split.

The score is built to miss little, so it will send some benign lesions for further work-up. That is the right trade for a triage tool in a general orthopaedic clinic, where the danger is a missed metastasis or sarcoma. It does not replace specialist imaging review or biopsy.

• Score an incidental lytic long-bone lesion systematically: margin, endosteal erosion, periosteal reaction, pathological fracture and soft-tissue mass.
• Ask every adult with a lytic lesion about a previous cancer; it is part of the score.
• Treat a high score as a reason for urgent work-up and referral, not a diagnosis.
• Expect some benign lesions to score as possibly malignant — specificity was about two-thirds.
• Use the benign-versus-malignant split when discussing with colleagues; the four tiers were less reproducible.

Why it matters

It gives general orthopaedic clinics a reproducible way to decide which incidental lytic lesions need urgent work-up.

Don't overread it

A single-centre retrospective validation; specificity was modest and the full score was only moderately reproducible.

The statistics, in plain English

Sensitivity of 97% means about 3 in 100 malignant lesions would be scored as possibly benign. Specificity of 67% means about a third of benign lesions would be flagged. A kappa of 0.52 means readers agreed moderately — better than chance, short of reliable — on the four-tier grade.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.