Abstract P2-20-08: Validation and comparison of breast graded prognostic assessment score and modified breast graded prognostic assessment in patients with brain metastases as prognostic tool

Background: Brain metastases (BM) occur in 15-35% of patients (p) with metastatic breast cancer (MBC), confering bad prognosis and the major limitation of life expectancy and quality of life. Nevertheless, there is a wide prognostic spectrum within this group of p. Clinical scores have been developed trying to classify according to its prognosis. Initially, the Breast Graded Prognostic Assessment (B-GPA, Sperduto et al, 2012) Then, it was adapted by adding the number of brain metastases, which is called modified Breast Graded Prognostic Assessment (mB-GPA, Subbiah et al 2015). We tried to validate these scores and compare them to determine which is the best prognostic predictor tool. Method: This is an ambispective study including all patients diagnosed of breast cancer’s brain metastases treated in a single cancer comprehensive center. Identification of patients and updates of follow up were performed through Radiation Oncology department registries and Pathology records. Both clinical and pathological data were collected. B-GPA and mB-GPA were calculated for each P. We analyze the survival after BM (SaBM) determining it as the time from the moment of diagnosis of the BM until the death due to the disease or any other cause. The Kaplan-Meier method was used to calculate SaBM and the univariate analyses was conducted using the Cox proportional hazard regression model. ROC curves were performed to compare both scores. Results: We included 169 P with a median age of 50 years (29-81). At last follow-up (April 2019) 90% of the P had died. Histological subtypes: 41.8% were luminal-like, 36.1% Her2 positive and 22% triple negative tumors. 25% of these P were diagnosed with stage IV. Nineteen percent of p presented only BM as the first site of relapse. A third of the P presented more than 3 brain metastases. WBRT was applied to 68.2% P, while stereotactic radiosurgery and surgery was performed in 18.2% and 18.2% respectively. The median SaBM was 12 months (m) (95% Confidence Interval (CI) [8.0-16.0m]. In the univariated analysis SaBM was longer in Her2 P than Luminal and TN P (26 m vs 11m and 8m p=0.01). Regarding the number of lesions, more than 3 present worse survival than p with 2-3 and only one metastasis (8m vs 11m vs 22m p=0.046). In the univariated analysis B-GPA and mGPA showed significant correlation with prognosis. Patients in group 4 of B-GPA (3.5-4 points) presented a better SaBM than the other groups, being statistically significant (see table 1) Likewise, group 4 of mB-GPA (3.5-4 points) showed a better stadistically significant survival (see table 1). The ROC curves showed that GPA and mGPA have very similar prognostic capabilities, perhaps slightly in favor of modified GPA (GPA: 0.269 and mGPA: 0.286) Conclusion: Our series validates the previous results regarding the GPA scores that allow discriminating risk groups with significant different survival. In our series mB-GPA was slightly better predictor of prognostic. In the univariate analysis B-GPA and mB-GPA, number of lesions and histological subtypes were the most important prognosis factors for MBC patients with BM. This situation can allow us to choose and better adjust the therapeutic effort in each patient, which has to be valued by a multidisciplinary team. Citation Format: Carles Fabregat Franco, Agostina Stradella, Valentin Navarro, Jennifer Linares, Sabela Recalde Penabad, Roser Velasco Fargas, Marta Simo Parra, Adela Fernandez Ortega, Andrea Vethencourt, Catalina Falo Zamora, Silvia Vazquez Fernandez, Rafael Villanueva Vazquez, Manuel Galdeano, Miguel Gil-Gil. Validation and comparison of breast graded prognostic assessment score and modified breast graded prognostic assessment in patients with brain metastases as prognostic tool [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P2-20-08.