Radiomics in the prediction of metal work failure in thoracolumbar spine fixations

Retrospective Cohort Study Objectives: To identify factors that could predict metalwork failure in thoracolumbar instrumentation for different indications.

A retrospective analysis was conducted on patients who underwent thoracolumbar fixation in a single unit between January 2014 to January 2023. Radiological parameters were measured using CT and MRI. The study was analysed in a per-case and a per-screw basis using T-test, Chi-Square test, Logistic regression and ROC analysis.

Over a 9-year period, 444 patients underwent 486 thoracolumbar instrumentation surgeries. Complications were observed in 20% of cases, with post-operative wound infection being the most common (9.7%). Metalwork failure was identified in 31 cases (6.38%). In the per-patient analysis, the presence of wound infection and average pedicle cancellous bone density (<280 HU), were statistically significant factors in predicting metalwork failure. Wound infection was a significant predictor for metal work failure in both degenerative and traumatic indications for spinal fixations. In the per-screw analysis, the screw-to-pedicle area ratio was significantly higher (>0.21) in screws without metalwork failure. For traumatic indications, the screw-to-pedicle area ratio (<0.25) and Charleston Comorbidity Index (CCI) (>0.15) were significant predictors. For degenerative indications, presence of wound infection and total pedicle bone density (<220 HU) were significant predictors. For neoplastic indications, only age (>66 years) was a predictor.

This study highlights the significance of avoiding post-operative wound infection, the use of screws with a larger diameter and higher pedicle cancellous bone density (>280 HU) in the reducing the risk of metalwork failure in thoracolumbar fixation.