Validation of the Chowdhury overall survival score in patients with melanoma brain metastasis treated with Gamma Knife Radiosurgery.

Melanoma brain metastases (MBM) are common in patients with stage IV disease. For Gamma Knife radiosurgery (GKRS) on MBM, risk scores such as RPA and melanoma-GPA aid to identify prognostic subgroups. This study aimed to validate the overall survival (OS) risk score developed by Chowdhury et al. in our center's patient cohort. A total of 104 MBM patients were treated with GKRS between 1/1/2002 and 31/12/2014 in our institution. Patients were categorized according to RPA, melanoma-GPA and Chowdhury OS score. The Kaplan-Meier method was used to estimate overall survival, and predicted survival probabilities were calculated for calibration. Cox proportional hazards regressions were performed to identify additional risk factors. Overall, median follow-up time was 80 months, while median OS (mOS) after GKRS was 6 months. Stratified according to the Chowdhury OS score, mOS in the high, medium and low risk group was 3.4, 7.1, and 10.0 months, respectively. The addition of other patient or disease characteristics to the Chowdhury OS model did not improve its performance. The C-index of the melanoma-GPA was 0.46 while the Chowdhury OS had an index of 0.67. In comparison with the RPA and melanoma-GPA, the Chowdhury OS score more accurately distinguished between separate risk groups among patients with MBM treated with GKRS. Contrary to the original study by Chowdhury, follow-up time was sufficient here for the low-risk group to reach the mOS time of 10 months.