Reliability of the Modified Tokuhashi Score for Decision-Making in Patients Surgically Treated for Thoracolumbar Spine Metastasis – Experience in a Single-Center

Author(s): Sofia J. M. Johansson MD, Benoît Maeder MD, Constantin Tuleasca MD-PhD, Juan Barges-Coll MD-MSc

Background: Oncological patients frequently suffer from spinal metastases. The decision to perform surgery mostly rely over scoring systems that could be outdated.

Methods: A retrospective study of 60 patients was performed, wherein different scoring methods were used and compared to evaluate their accuracy in predicting survival for patients who underwent spinal surgery. Patient scoring systems were assessed by ROC curve, Kaplan-Meier, and Cox proportional hazard regression analyses.

Results: The performance of the modified Tokuhashi score was only marginally better than that of a random classifier, but Kaplan-Meier curves of the Tomita and modified Tokuhashi scores exhibited statistically significant differences between different score groups according to the logrank test. The univariate analysis revealed the statistical significance of a small set of factors for predicting patient survival, and the multivariate analysis revealed the statistical significance of an even smaller set of factors. Neither prior nor ongoing immuno/hormonotherapy significantly improved patient survival, but patients with prostate or breast primary tumors nonetheless appeared to have a slightly better likelihood of longterm survival.

Conclusion: In total, patient scoring systems have only demonstrated mediocre reliability for predicting survival following spinal surgery. The overall heterogeneity of patient profiles requires nuanced and tailored decision-making, and current scoring systems can at most provide very weak guidance in this process.