Prediction of early distant brain failure after stereotactic radiotherapy for brain metastases.

BACKGROUND

Stereotactic radiotherapy (SRT) is the main treatment for patients with 1 to 3-5 brain metastases (BMs) but with the compromise of a higher risk of distant brain failure (DBF) in comparison with whole brain radiotherapy. Early DBF has a significant negative impact on overall survival (OS). We propose to build and validate a prediction model of early DBF.

METHODS

Early DBF was defined as the appearance of unknown BMs on the first post-SRT magnetic resonance imaging (MRI). A nomogram was built for the prediction of early DBF using patients from a cohort of 537 SRT courses. The nomogram was evaluated for early DBF classification using the chi-squared test. Prediction of DBF-free survival and OS was tested using Kaplan-Meier curves and the log-rank test. The model was validated on an external cohort of 160 subsequently delivered SRT courses.

RESULTS

In the cohort of 537 SRT courses, early DBF occurred in 17% cases. The number of BMs and the DS-GPA score were significant predictors of early DBF. The nomogram demonstrated robust performances for early DBF classification (χ 23.7,  < .0001), DBF-free survival (χ 35.5  < .0001, Figure 1), and OS (χ 38.9,  < .0001). On the validation cohort, the same nomogram achieved a χ of 15.3,  = .0005 for DBF-free survival and a χ of 8.6,  = .01 for OS.

CONCLUSION

Our study provides a robust predictive model for early DBF, validated in an independent cohort. This nomogram could improve clinical outcomes and treatment personalization.