Systematic review of 99 extremity bone malignancy survival prediction models

BACKGROUND: Various prediction models have been developed for extremity metastasis and sarcoma. This systematic review aims to evaluate extremity metastasis and sarcoma models using the utility prediction model (UPM) evaluation framework. METHODS: We followed Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and systematically searched PubMed, Embase, and Cochrane to identify articles presenting original prediction models with 1-year survival outcome for extremity metastasis and 5-year survival outcome for sarcoma. Identified models were assessed using the UPM score (0-16), categorized as excellent (12-16), good (7-11), fair (3-6), or poor (0-2). A total of 5 extremity metastasis and 94 sarcoma models met inclusion criteria and were analyzed for design, validation, and performance. RESULTS: We assessed 5 models for extremity metastasis and 94 models for sarcoma. Only 4 out of 99 (4%) models achieved excellence, 1 from extremity metastasis and 3 from sarcoma. The majority were rated good (62%; 61/99), followed by fair (31%, 31/99) and poor (3%, 3/99). CONCLUSIONS: Most predictive models for extremity metastasis and sarcoma fall short of UPM excellence. Suboptimal study design, limited external validation, and the infrequent availability of web-based calculators are main drawbacks. LEVEL OF EVIDENCE: This study is classified as Level 2a evidence according to the Oxford 2011 Levels of Evidence. Trial registration This study was registered in PROSEPRO (CRD42022373391, https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=373391 ).