No abstract

No abstract

No abstract

Long bone pathological fractures very much reflect bone metastases morbidity in many types of cancer. Bearing in mind that they not only compromise patient function but also survival, identifying impending fractures before the actual event is one of the main concerns for tumor boards. Indeed, timely prophylactic surgery has been demonstrated to increase patient quality of life as well as survival. However, early surgery for long bone metastases remains controversial as the current fracture risk assessment tools lack accuracy. This review first focuses on the gold standard Mirels rating system. It then explores other unique imaging thresholds such as axial or circumferential cortical involvement and the merits of nuclear imaging tools. To overcome the lack of specificity, other fracture prediction strategies have focused on biomechanical models based on quantitative computed tomography (CT): computed tomography rigidity analysis (CT-RA) and finite element analysis (CT-FEA). Despite their higher specificities in impending fracture assessment, their limited availability, along with a need for standardization, have limited their use in everyday practice. Currently, the prediction of long bone pathologic fractures is a multifactorial process. In this regard, machine learning could potentially be of value by taking into account clinical survival prediction as well as clinical and improved CT-RA/FEA data.