When to refer for surgery in long bone metastases: A powered multivariate analysis of fracture risk and simple algorithm for referral for surgery (the metastatic fracture prediction met FRAP score)

BACKGROUND: It is difficult to predict the risk of pathological fracture in patients with bone metastases. Popular scoring systems like Mirels' may overestimate need for surgery by 20%. The aim was to identify predictors of fracture at 6 months in patients with long bone metastases managed non-surgically. PATIENTS AND METHODS: This was an adequately powered retrospective analysis of individuals with a long bone metastasis. Patients were identified from radiology reports and excluded if they had prophylactic surgery or died within 3 months. A literature review and pilot study identified 13 variables of interest. Multivariate analysis was used to generate the metastatic fracture prediction (Met FRAP) score. RESULTS: From 2009 to 2021 (13 years), 2142 patients with new long bone metastases were identified. 1577 were excluded, leaving a dataset of 565 lesions in 503 patients managed non-surgically. The cohort were 45% female with a mean age of 71 years. Median follow-up in living patients was 2 years (range 1-12 years). The metastatic fracture rate at 1 year was 26%. On multivariate analysis, independent predictors of metastatic fracture included cognitive impairment, metastatic load (p < 0.001 for both), primary cancer type (p < 0.01), radiotherapy, pain requiring opiates and radiographic appearance (p < 0.05 for all). These were used to generate the Met FRAP score. Met FRAP score predicted metastatic fracture at 6 months with 82% accuracy and was more sensitive and specific than Mirels score (61% accuracy, p < 0.001). DISCUSSION: The Met FRAP score accurately predicts metastatic fracture at 6 months and appears to be more reliable than Mirels score.