Does Graded Prognostic Assessment outperform Recursive Partitioning Analysis in patients with moderate prognosis brain metastases?

Abstract

Aim: To compare the clinical utility of the Recursive Partitioning Analysis (RPA) and Graded Prognostic Assessment (GPA) in predicting outcomes for moderate prognosis patients with brain metastases. Methods & materials: We reviewed 101 whole brain radiotherapy cases. RPA and GPA were calculated. Overall survival was compared. Results: Sixty-eight patients had moderate prognosis. RPA patient characteristics for increased death hazard were ≤10 WBRT fractions or no surgery/radiosurgery. GPA patients had increased death risk with no surgery/radiosurgery or lower Karnofsky Performance Status. Conclusion: The indices have similar predicted survival. Patients scored by RPA with longer radiation schedules had longer survival; patients scored by GPA did not. This indicates GPA is more clinically useful, leaving less room for subjective treatment choices.