Original Article
Validation of Modified Breast Graded Prognostic Assessment for breast cancer patients with brain metastases: extra-cranial disease progression is an independent risk factor
Abstract
Background: Breast cancer (BC) patients with brain metastases (BM) are heterogeneous with markedly variable survival. The Breast Graded Prognostic Assessment (B-GPA) and Modified B-GPA (mB-GPA) have been proposed as useful tools for stratifying survival in this population. However, extra-cranial disease progression, a clinically important variable, is not incorporated into the final model. We undertook the validation of B-GPA and mB-GPA in an Asian cohort and further explore extra-cranial disease progression as a prognostic factor.
Methods: Data of BC patients with newly diagnosed BM between 2006 and 2017 was extracted retrospectively from a prospectively maintained institutional database. Patients were classified based on their B-GPA and mB-GPA scores. Univariate (UVA) and multivariate analysis (MVA) using the Cox proportional hazard model were performed to investigate the factors prognostic of overall survival (OS). The Kaplan-Meier method was used to estimate OS and log-rank test to compare survival between scores. The performances of B-GPA and mB-GPA were compared using Harrell’s concordance index (C-index) and Akaike information criterion (AIC).
Results: In our cohort of 282 patients, the B-GPA and mB-GPA were validated as prognostic tools for OS, demonstrating excellent separation between survival curves (P<0.001). In MVA, we found all components of mB-GPA (age, performance status, number of BM, tumour subtype) to be independent predictors of survival. C-index was 0.64 and AIC was 2,483.39 for B-GPA. mB-GPA demonstrated marginally better discrimination with a C-index of 0.65 and AIC of 2,445.78. Extra-cranial progression was an independent predictor for survival in our population (P<0.001).
Conclusions: The mB-GPA incorporates four simple clinical variables each of independent prognostic significance. Both B-GPA and mB-GPA demonstrate moderate discriminative capabilities for OS with mB-GPA performing marginally better. Inclusion of extra-cranial disease progression as a factor in future model development may further improve its prognostic value.
Methods: Data of BC patients with newly diagnosed BM between 2006 and 2017 was extracted retrospectively from a prospectively maintained institutional database. Patients were classified based on their B-GPA and mB-GPA scores. Univariate (UVA) and multivariate analysis (MVA) using the Cox proportional hazard model were performed to investigate the factors prognostic of overall survival (OS). The Kaplan-Meier method was used to estimate OS and log-rank test to compare survival between scores. The performances of B-GPA and mB-GPA were compared using Harrell’s concordance index (C-index) and Akaike information criterion (AIC).
Results: In our cohort of 282 patients, the B-GPA and mB-GPA were validated as prognostic tools for OS, demonstrating excellent separation between survival curves (P<0.001). In MVA, we found all components of mB-GPA (age, performance status, number of BM, tumour subtype) to be independent predictors of survival. C-index was 0.64 and AIC was 2,483.39 for B-GPA. mB-GPA demonstrated marginally better discrimination with a C-index of 0.65 and AIC of 2,445.78. Extra-cranial progression was an independent predictor for survival in our population (P<0.001).
Conclusions: The mB-GPA incorporates four simple clinical variables each of independent prognostic significance. Both B-GPA and mB-GPA demonstrate moderate discriminative capabilities for OS with mB-GPA performing marginally better. Inclusion of extra-cranial disease progression as a factor in future model development may further improve its prognostic value.
