Actual versus Predicted Overall Survival in Patients Treated Surgically for Metastatic Spinal Cord Compression

Abstract

Objective: To compare the predicted survivals based on Tokuhashi and Tomita scores, as well as clinical estimates by the oncologist, with the actual observed survival

Study Design: A retrospective observational study
Place and Duration of Study: This study was conducted at the Queen Elizabeth Hospital, Birmingham, UK between 20th June 2025 to 19th October 2025.

Methods: This was a retrospective observational study was carried out at Queen Elizabeth Hospital, Birmingham, UK between 20th June 2025 to 19th October 2025 of 32 patients treated surgically for metastatic spinal cord compression.

Results: A Cox proportional hazards model found Tokuhashi and Tomita scores to have negative predictive values. Log-rank tests demonstrated p-values <0.05 in the comparison of outcomes with Tokuhashi, Tomita and oncological expert predictions, indicating significant differences between predicted and actual survival. Spearman’s correlation coefficient showed a moderate positive correlation between Tokuhashi predictions and observed survival (p= 0.017), while Tomita predictions showed no correlation with observed survival (p = 0.893). The discriminatory predictive performance of the two scoring systems revealed areas under the curve of 0.57 and 0.369 for Tomita and Tokuhashi, respectively. Both showed slight non-specific agreement with actual survival outcomes, with kappa values of 0.04 and 0.07, respectively.

Conclusion: While both of these scoring system prognostic models can serve as guides to likely outcomes, they lacked predictive accuracy in our cohort. We recommend the development of more robust, individualised survival prediction tools for preoperative treatment planning in MSCC patients.