Objective: Patients with brain metastases were analyzed retrospectively to assess the risks and benefits of surgery with modern neurosurgical techniques, including image guidance coupled as indicated with corticography.
Methods: We retrospectively analyzed charts of patients treated surgically for brain metastases. We identified patients with single or multiple brain metastases who underwent craniotomies to reverse associated neurological symptoms or establish a diagnosis. We assessed patients according to recursive partitioning analysis (RPA) prognostic groups as well as functional grades of tumor location (eloquent versus noneloquent, Grades I-III). Perioperative complications, neurological outcomes after surgery, survival, and prognostic factors were analyzed. Statistical analysis of survival was performed with the Kaplan-Meier method. A P value of <0.05 was considered statistically significant.
Results: Two hundred eight patients were treated between March 1995 and December 2002. Patient age ranged from 31 to 82 years (median, 59 yr). One lesion was resected in 191 patients, and of 76 patients with multiple lesions, two or more metastases were resected in 17 patients. Tumors were located in eloquent cortex in 27 patients and near eloquent cortex in 124 patients. Four patients died within 30 days after surgery for a mortality rate of 1.9%. Neurological deterioration was noted in 13 patients (6%) after surgery for Grade I and II tumors and in 5 patients (19%) of 27 patients with Grade III tumors. Karnofsky Performance Scale scores were improved (68 patients) or unchanged (124 patients) in 192 patients and worse in 16 patients after surgery. The median survival time (MST) from the date of surgery was 8 months for all patients and 9 months for 163 patients who did not undergo prior whole-brain radiation therapy. There was no difference in survival between patients operated for single metastasis (MST, 8 mo) versus patients with two or three metastases (MST, 9 mo; P = 0.9364). By both univariate and multivariate analysis, variables significantly affecting outcome included a high Karnofsky Performance Scale score and RPA Class I assignment. By univariate analysis, significant treatment variables included postoperative radiotherapy and postoperative chemotherapy. The MSTs of RPA Class I, II, and III patients were 16.1 months, 7.2 months, and 1.4 months, respectively (P < 0.001, log-rank test). These survival data compare favorably with the stereotactic radiosurgery boost arm of the recently published Radiation Therapy Oncology Group 9508 trial.
Conclusion: In most patients with single or multiple brain metastases, surgical resection reversed or stabilized neurological symptoms with therapeutic benefit, conveying a notable survival advantage without apparent increased risk, particularly in RPA Class I patients. In patients with Grade III single metastasis or RPA Class II multiple metastasis, surgical judgment should be exercised, and stereotactic radiosurgery boost treatment may be preferable. An algorithm for treatment of brain metastases is proposed.