Preoperative versus postoperative single-fraction or fractionated stereotactic radiosurgery for resected brain metastases: a systematic review and network meta-analysis

Abstract

The optimal timing and fractionation of perioperative stereotactic radiotherapy for resected brain metastases remain uncertain. We evaluated the comparative evidence for preoperative and postoperative stereotactic treatment and the validity of indirect comparisons. We searched PubMed/MEDLINE, CENTRAL, Web of Science Core Collection, and ClinicalTrials.gov for comparative studies published or posted through 12 August 2026. Primary outcomes were local control or failure, leptomeningeal disease (LMD), and radiation necrosis (RN). Compatibility for quantitative synthesis was assessed by treatment definition, analysis unit, outcome definition, effect measure, and estimation method. Compatible patient-level hazard ratios (HRs) for local failure were combined in an exploratory fixed-effect network model. Incompatible evidence and patient-level and lesion-level estimates were synthesized separately. Risk of bias and certainty of evidence were also assessed. Sixteen comparative studies were included, comprising six randomized trials and ten non-randomized studies. Three studies provided compatible patient-level HRs and formed a four-treatment local-failure network. Postoperative single-fraction stereotactic radiosurgery (SRS) reduced local failure compared with observation in one randomized trial (HR, 0.46; 95% CI, 0.24–0.88). Estimates for preoperative versus postoperative single-fraction SRS (HR, 1.57; 95% CI, 0.38–6.49) and postoperative single-fraction SRS versus three- to five-fraction treatment (HR, 0.82; 95% CI, 0.09–7.32) were imprecise. All indirect comparisons had wide confidence intervals that included 1. Each direct comparison was informed by one study, and the network contained no closed loop. Some indirect estimates were sensitive to the evidence selected for the single-fraction versus fractionated-treatment comparison. Evidence for LMD and RN did not form compatible connected networks and was synthesized descriptively. Certainty was moderate for postoperative single-fraction SRS versus observation and very low for the remaining local-failure comparisons. Randomized evidence supports postoperative single-fraction SRS over observation for reducing local failure. However, current evidence does not establish a reliable hierarchy among preoperative SRS, postoperative single-fraction SRS, and postoperative fractionated treatment. Evidence for LMD and RN remains insufficient for network comparison. Treatment decisions should account for individual clinical characteristics until adequately powered prospective comparisons become available. https://www.crd.york.ac.uk/PROSPERO/ , identifier CRD420261341468.