Evaluation of Scoring Systems and Prognostic Factors in Patients With Spinal Metastases From Lung Cancer.

STUDY DESIGN A retrospective study of 180 patients with lung cancer spinal metastases, wherein prognostic score-predicted survival was compared with actual survival. OBJECTIVE To evaluate and compare the accuracy of prognostic scoring systems in lung cancer spinal metastases. SUMMARY OF BACKGROUND DATA The modified Tokuhashi, Tomita, modified Bauer, and Oswestry scores are currently used to guide decisions regarding operative treatment of patients with spinal metastases. The best system for predicting survival in patients with lung cancer spinal metastases remains undetermined. The high incidence of spinal metastases from lung cancer and improved survival of patients treated with systemic therapy warrants evaluation of these scoring systems in this particular context. METHODS Patients with lung cancer spinal metastases treated at our institution between May 2001 and August 2012 were studied. Fifty-one patients were treated surgically. The primary outcome measure was survival from the time of diagnosis. Scoring-predicted survival was compared with actual survival. Potential prognostic factors were investigated using Cox regression analyses. Predictive values of each scoring system for 3- and 6-month survival were measured via receiver operating characteristic (ROC) curves. RESULTS Histological subtype (P = 0.015), sex (P = 0.001), Karnofsky performance scale (P = 0.001), extent of neurological palsy (P = 0.002), and visceral metastases (P = 0.037) are significant predictors of survival. Besides the Oswestry spinal risk index, no significant differences were found between different prognostic subgroups within the individual scoring systems. Although the modified Bauer score was most accurate, all four scoring systems had areas under the ROC curve 0.5 or less. CONCLUSION Although better prognostic scores correlated with longer survival, all four scoring systems are inaccurate in prognosticating patients with lung cancer spinal metastases. Specific lung cancer histology appears prognostic and should be considered, especially given the increased survival of patients receiving new targeted therapies appropriate to their disease. LEVEL OF EVIDENCE 3.