A Competing Risks Analysis Assessing Predictors Of Neurologic And Non-Neurologic Death In Patients With Brain Metastasis Initially Treated With Upfront Stereotactic Radiosurgery Without Whole-Brain Radiation Therapy

Recent clinical trials suggest rates of neurologic death are higher in patients with brain metastasis initially treated with stereotactic radiosurgery (SRS) alone as compared to those treated with SRS, plus subsequent whole-brain radiation therapy (WBRT). In this study we attempt to discern the factors predictive of neurologic death in patients with brain metastasis treated with upfront SRS alone. Six hundred thirty-seven of 738 patients at an academic center were treated initially with SRS and were either alive or had complete information regarding cause of death at the time of analysis. Neurologic death was defined as death from progressive neurologic dysfunction irrespective of systemic disease status. Univariate analysis identified predictive covariates with a P value ≤ .1 for consideration of inclusion into a multivariate (MVA) Cox proportional hazards model. Competing risks analysis was performed to estimate the subdistribution hazard ratios (HRs) for neurologic death and non-neurologic death for each predictor. Of 738 patients treated with upfront SRS alone, 636 (86.2%) patients were deceased at time of analysis. Of these patients, the cause of death was neurologic in 226 (35.5%), non-neurologic in 309 (48.6%), and unknown in 101 (15.9%). Predictors of neurologic death identified on univariate analysis included histology (P = .03), minimal SRS dose (P = .03), number of brain metastases (P < .0001), systemic disease status (P = .003), KPS (P < .0001), and post-SRS RTOG symptom grade (P < .0001). Multivariate competing risks analysis identified an increased hazard of neurologic death in patients with a higher post-SRS RTOG symptom grade (P < 0.0001) of 2 (HR = 1.9, 95% CI 1.3-2.7), 3 (HR = 3.1, 95% CI = 1.8-5.4) or 4 (HR = 5.8, 95% CI 3.2-10.3), and melanoma histology (HR = 2.0, 95% CI 1.1-3.6, P = .03). Progressive systemic disease was associated with an increased hazard of non-neurologic death (HR = 1.8, 95% CI 1.4-2.4, P = .0001). In patients with brain metastasis treated upfront with SRS alone, higher post-SRS RTOG symptom grade was the dominant factor in predicting neurologic death. This finding validates the historic notion that metastatic patients with symptomatic intracranial disease have an inferior prognosis secondary to an increased hazard of neurologic death. Conversely, the presence of progressive systemic disease was the primary predictor of non-neurologic death. Careful consideration of the effect of classic prognostic factors on terminal events may be useful for defining high-risk disease.