OBJECTIVETo evaluate the predictive values of Tokuhashi score, revised Tokuhashi score and Tomita score systems for life expectancy and treatment options in patients with spinal metastasis.

METHODSFrom February 1996 to January 2009, spinal operations in 104 cases with spinal metastasis were performed in our hospital. There were 65 males and 39 females, with an average of 53.4 years (median 52.5 years). To calculate AUC (area under the curve) values of Receiver Operating Characteristic (ROC) curves of three scores, and to analyze the accuracy of prediction of life expectancy. To compare the actual survival time with the expected survival time of the three scores by Kaplan-Meier method. Spearman correlation analysis was performed between the survival time and three scoring systems.

RESULTSAll cases were followed-up with an average duration of 10.9 months, and 77 patients died. AUC analysis of ROC curves showed that the difference of the accuracy of the three scores was not significant. AUC in all groups of Tokuhashi Score was low, with a poor diagnostic accuracy. In the "died within 3 months" and "died within 6 months" groups of revised Tokuhashi score, the accuracy was low, while high in the other two groups. The AUC values of Tomita score in "died within 6 months" and "died within 24 months" were high, with a great diagnostic accuracy while the other two groups were low with a low diagnostic accuracy. Kaplan-Meier survival curve analysis showed that the actual survival time in all three scores was not entirely consistent with the expected survival time. Tokuhashi score and revised score were positively correlated with the survival time while that of Tomita score was negative.

CONCLUSIONAll the three prognosis scores in patients with spinal metastasis were closely related with survival time. The combination of Tokuhashi score and Tomita score may be applied to better predict postoperative survival prognosis and guide the surgical options for patients with spinal metastasis.