The longitudinal risk of hemorrhage of melanoma brain metastases after Gamma Knife radiosurgery

OBJECTIVE The objective of this study was to analyze the hemorrhagic risk of melanoma brain metastases after Gamma Knife radiosurgery (GKRS). METHODS A prospective institutional database was retrospectively queried to identify patients who underwent GKRS for melanoma brain metastases between 1990 and 2021. Lesional hemorrhage was defined as definite or possible based on radiologists’ readings, and severity was graded according to Common Terminology Criteria for Adverse Events. RESULTS Two hundred ninety-one patients with 1083 lesions treated in 419 sessions were identified. The mean (± SD) patient age was 60 ± 15 years, and 61% were male. The median follow-up period for overall survival (OS) was 11 (range 0–214) months with 581 patient-years. Definite/possible lesional hemorrhages occurred in 13% of lesions, with grade 3 hemorrhages observed in 4% of lesions. Surgical intervention was required in 2% of cases (5% of patients), and all resected lesions were pathologically consistent with melanoma. A decreased risk of definite/possible lesional hemorrhage was associated with a later time period between 2015 and 2021 (OR 0.45, 95% CI 0.266–0.75, p = 0.0021), increased marginal dose (OR 0.91, 95% CI 0.83–0.99, p = 0.037), antiplatelet use post-GKRS (OR 0.195, 95% CI 0.083–0.46, p < 0.001), and whole-brain radiotherapy (WBRT; OR 0.53, 95% CI 0.344–0.82, p = 0.0042). After 2015, more patients received anticoagulation, B-Raf proto-oncogene inhibitors, and immune checkpoint inhibitors, and fewer received bevacizumab (p < 0.001). The cumulative risk of lesional hemorrhage was 17%–20% at 36 months from GKRS, with 95%–96% of cases occurring within 12 months. The median patient OS was 11 (95% CI 9–13) months, and multivariate Cox regression analysis revealed that antiplatelet agents (hazard ratio [HR] 0.66, 95% CI 0.45–0.96, p = 0.031) and immune checkpoint inhibitors (HR 0.35, 95% CI 0.26–0.48, p < 0.001) were associated with longer OS, while WBRT (HR 1.36, 95% CI 1.02–1.81, p = 0.037) and definite/possible hemorrhage (HR 1.39, 95% CI 1.04–1.85, p = 0.024) were associated with shorter OS. CONCLUSIONS The definite hemorrhage risk of melanoma brain metastases after GKRS was 17% in the first 3 years and 95% of the lesional hemorrhage occurred within the 1st year. Surgical intervention was needed in 5% of patients. Antiplatelet agents and immune checkpoint inhibitors were associated with improved OS, while definite/possible hemorrhage was associated with worse OS.