METASTATIC DISEASE IN THE SPINE

Abstract

Objective: In 1989 Mirels published a scoring system for identifying impending pathological fractures in long bones. However, the spine is the most common site of skeletal metastases. A MR-based scoring system is proposed to quantify the risk of sustaining a pathological fracture through a metastatic lesion in a vertebral body.

Methods: A retrospective analysis of 101 vertebral body metastatic lesions was carried out. The metastases were identified through the onco-radiology database. Only lesions with a MR scan and subsequent imaging within 24-months of the index scan were included. Variables potentially predictive of impending fracture were analysed for significance. The significant variables were then statistically weighted. The original MR scans were scored, and the subsequent imaging was used to identify which lesions fractured. The scores were compared between the fracture and non-fracture group. Analysis was carried out for each predictive variable to establish whether they were individually as good as the scoring system alone in predicting fracture. Intra and inter-observer variability was assessed using kappa statistics.

Results: Twenty-one of the 101 lesions fractured within 24 months. A mean score of 0.65 was identified in the non-fracture group, whilst the fracture group had a mean score of 6.52 (p< 0.0001). The percentage risk of a lesion sustaining a pathological fracture was calculated for any given score. As the score increased above 4, so did the percentage risk of fracture (sensitivity 85.7%, specificity 97.5%). Very good intra and inter-observer agreement was present, showing the scoring system to be reliably reproducible.

Conclusions: The authors propose that all painful vertebral body metastatic lesions be evaluated by MR scanning. Lesions with a score of 3 or less can be left untreated. Lesions with scores of 4 or higher are at risk of fracture and should be considered for prophylactic cement augmentation.