BackgroundBrain metastases (BM) from non-small cell lung cancer (NSCLC) are the most frequent intracranial tumors. To identify patients who might benefit from intracranial surgery, we compared the six existing prognostic indexes(PIs) and built a nomogram to predict the survival for NSCLC with BM before they intended to receive total intracranial resection in China.MethodsFirst, clinical data of NSCLC presenting with BM were retrospectively reviewed. All of the patients had received total intracranial resection and were randomly distributed to developing cohort and validation cohort by 2:1. Second, we stratified the cohort using a recursive partitioning analysis(RPA), a score index for radiosurgery (SIR), a basic score for BM (BS-BM), a Golden Grading System (GGS), a disease-specific graded prognostic assessment (DS-GPA) and by NSCLC-RADES. The predictive power of the six PIs was assessed using the Kaplan–Meier method and the log-rank test. Third, univariate and multivariate analysis were explored, and the nomogram predicting survival of BMs from NSCLC was constructed using R 3.2.3 software. The concordance index (C-index) was calculated to evaluate the discriminatory power of the nomogram in the developing cohort and validation cohort.ResultsBS-BM could better predict survival of patients before intracranial surgery compared with other PIs. In the final multivariate analysis, KPS at diagnosis of BM, metachronous or synchronous BM and the histology of lung cancer appeared to be the independent prognostic predictors for survival. The C-index in the developing cohort and validation cohort were 0.75 and 0.71 respectively, which was better than the C-index of the other six PIs.ConclusionsThe new nomogram is a promising tool in further choosing the candidates for intracranial surgery among NSCLC with BM and in helping physicians tailor suitable treatment options before operation in clinical practice.