P11.05.A EXTERNAL VALIDATION OF THE LUNG-MOLGPA TO PREDICT SURVIVAL AFTER STEREOTACTIC RADIOSURGERY FOR BRAIN METASTASES OF NON-SMALL CELL LUNG CANCER

Abstract

Abstract BACKGROUND Accurate prediction of survival in patients with brain metastases is important for shared decision making. An updated version of the Lung-molGPA prognostic index estimates survival in patients with brain metastases of non-small cell lung cancer (NSCLC). This version includes programmed death ligand 1 (PDL1) expression as a prognostic factor. As this version has not yet been externally validated, we investigated whether the Lung-molGPA accurately predicts survival in a cohort of patients treated with stereotactic radiosurgery (SRS) for brain metastases. MATERIAL AND METHODS All patients (n = 140) treated with SRS for brain metastases of NSCLC (adenocarcinoma) between 2015 and 2020 were included in this retrospective study. GPA score was calculated for each patient based on: Karnofsky Performance Score (0 points for ≤70, 0.5 for 80, 1.0 for 90-100), age (0 points for ≥ 70 years, 0.5 for < 70), number of brain metastases (0 points for ≥ 5, 0.5 for < 5), extracranial metastases (0 points for present, 1.0 for absent), expression of either epidermal growth factor receptor (EGFR) or anaplastic lymphoma kinase (ALK) (0 points for negative or unknown, 0.5 for positive in either), and lastly expression of PDL1 (0 points for < 1% or unknown, 0.5 for ≥ 1%). A higher GPA score predicts improved survival. Kaplan-Meier analysis evaluated overall survival and hazard ratios were estimated by a Cox regression model. Predictive performance was evaluated using discrimination (C-statistic or area under the curve) and calibration (Hosmer-Lemeshow test). RESULTS Median age at diagnosis was 64 years (range 37-83) and median survival was 13 months. The cohort was divided into four prognostic groups, with a higher GPA score corresponding with favorable survival. The proportion of patients into each prognostic group was 12% (n = 17) with a GPA score of 0-1.0, 39% (n = 54) with a GPA score of 1.5-2.0, 43% (n = 59) with a GPA score of 2.5-3.0, and 5% (n = 8) with a GPA score of 3.5-4.0. Patients with a higher GPA score had better overall survival compared with lower GPA scores (p-values < 0.05 for all pairwise comparisons except for the most favorable prognostic group). Predicted and observed median survival were: 6 versus 7 months (GPA score 0-1.0), 15 versus 11 months (GPA score 1.5-2.0), 30 versus 30 months (GPA score 2.5-3.0), and 52 versus 24 months (GPA score 3.5-4.0). In the external validation, the GPA had a C-statistic of 0.70 (95% CI 0.59-0.82) and passed the Hosmer-Lemeshow goodness of fit test (p-value = 0.55). CONCLUSION The Lung-molGPA was externally validated in our population. The Lung-molGPA accurately predicted survival in our cohort of patients treated with SRS for brain metastases of NSCLC (adenocarcinoma) except for the small proportion of patients in the most favorable prognostic group. This prognostic tool is useful for counseling of patients in clinical practice.