Phase 3 international trial of adjuvant whole brain radiotherapy (WBRT) or observation (Obs) following local treatment of 1-3 melanoma brain metastases (MBMs).

9500 Background: The role of adjuvant WBRT in MBMs is controversial. This trial compares WBRT with Obs after local treatment of 1-3 MBMs. Methods: The primary endpoint is distant intracranial failure (DIF) within 12 months of randomization. The a priori neurocognitive function (NCF) endpoint is Hopkins Verbal Learning Test-Revised (HVLT-R) delayed recall at 4 months. Secondary endpoints include local failure (LF), overall survival (OS) and global quality of life (QoL). Analyses were conducted on intention-to-treat basis with nominal two-sided significance level 5%. Drug therapy was allowed. Effective drugs became available during trial and their impact was analysed. Results: Of 586 eligible patients (pts), 215 consented from 31 sites in 3 countries (Australia, UK and Norway) between 2009 and 2017. Eight (0.04%) who withdrew or had no data collected were excluded. 107 randomized to Obs and 100 to WBRT. Mean age 62 years, 67% males, 61% with single MBM of mean size 2cm, 67% had extracranial disease at randomization. The two arms were well matched. NCF was completed by English speakers; 50 WBRT and 70 Obs at baseline, declining to 26 and 35 respectively at 4 months. Within 12 months, 54 (50.5%) Obs had DIF compared with 42 (42.0%) WBRT pts (OR 0.71; 95%CI 0.41-1.23; p = 0.222). There was no difference in LF (p = 0.100) or OS (log-rank p = 0.861). 53% (Obs) and 59% (WBRT) pts were alive at 12 months. There was no significant between-group difference in mean intervention effect on global QoL (p = 0.083). Pts who received T-cell checkpoint inhibitors and/or mitogen-activated protein kinase (MAPK) pathway inhibitors and WBRT before or within 12 months of randomization had DIF rate 29% compared with Obs and no systemic therapy had 44%, but was not significant (p = 0.228). Obs had greater relative improvement from baseline in HVLT-R at every timepoint. At 4 months, Obs had 20.9% improvement from baseline in HVLT-R-delayed recall compared to 2.7% decline in WBRT; overall adjusted average intervention effect 23.6% (95%CI 9.0, 38.2; p = 0.0018). There was no difference in time to cognitive failure or in proportions with global cognitive impairment. Conclusion: This level one evidence shows WBRT does not improve outcomes in MBMs. This practice-changing trial justifies the recent move away from WBRT that occurred during the course of the trial. Clinical trial information: NCT01503827.