Which prognostic index is most appropriate in the setting of delayed stereotactic radiosurgery for brain metastases?

Objectives: To determine if five commonly used prognostic indices (PIs) - recursive partitioning analysis (RPA), Score Index for Radiosurgery (SIR), Basic Score for Brain Metastases (BSBM), graded prognostic assessment (GPA), and the diagnosis-specific GPA - are valid following delay between diagnosis and treatment of brain metastases. Methods: In a single-institutional cohort, records of patients who underwent stereotactic radiosurgery (SRS) more than 30 days from diagnosis of brain metastases were collected, and five PI scores were calculated for each patient. For each PI, three score-based groupings were made to examine survival differences by means of adjusted log-rank analysis and area under the curve (AUC). Results: Of 121 patients with sufficient PI information, 72 underwent SRS more than 30 days after diagnosis. Median age and Karnofsky performance status were 60 years and 80, respectively. Forty-three (60%) patients had lung primaries. Prior to SRS, 38 (52.8%) and 12 (16.7%) patients underwent whole brain radiation therapy (WBRT) and surgery, respectively. Two (2.8%) patients underwent both WBRT and surgery prior to SRS. A median of two lesions were treated per SRS course. Median survival of the cohort was 9.0 months. Using adjusted log-rank analysis for pairwise comparison, BSBM and GPA showed significance between two out of the three prognostic groups, while the other scores showed either one or no significant differences on comparison. AUC demonstrated good applicability for BSBM, RPA, and GPA, although SIR was statistically less prognostic than the other PIs. Conclusion: The PIs analyzed in this study were applicable in the setting of delayed SRS. Although these data are hypothesis generating, they serve to encourage further analyses to validate a PI that is most optimal for these patients.