Determination of Tumor Gene Expression in Non–small Cell Lung Cancer Patients Who Develop Brain Metastasis Diagnosis: Analysis of a Prospective Trial

Abstract

Brain metastasis (BM) occur in 30-50% of patients with non-small cell lung cancer (NSCLC). If BM are discovered after development of symptoms, patients have worse outcomes. Biomarkers for the development of BM are lacking. The current series is a genomic analysis of a completed prospective trial in which gene expression profiling was performed on RNA samples from the time of diagnosis of NSCLC. Between 2011 and 2015, a total of 101 patients with histologically-confirmed NSCLC were prospectively enrolled onto a clinical trial in which genomic analysis was performed on the RNA from a diagnostic FNA. RNA was extracted from a single pass specimen after the diagnostic biopsies were obtained. RNA from FNA were processed for gene expression analysis, data normalization, and quality control. Classification algorithms training on frozen FNAs was used. Retrospective analysis of BM-related outcomes was performed using patient electronic medical records including: date of BM development, any metastasis, and last follow-up. A competing risk analysis was performed for the effects of the prognostic factors on the risk of developing BM vs. other metastasis. Cumulative incidence functions of BM types in each subgroups were estimated for every prognostic factor. Gray’s test was used to assess statistical significance. Genes were selected through previous analysis of microarray data sets used for training and testing. Normalization and batch correction methods were applied. Gene expression signatures were made for histology using a classification algorithm using gene array tools. A total of 49 of 101 patients died at the time of last follow-up. Median time of follow-up was 12.98 months (range 0.23 to 66.53 months). A total of 17 patients developed BM; risk of brain met was lower than 50% ( 1 – 0.842 = 0.158), so no median survival time is available. Median survival time after diagnosis of BM was 3.581 months (95% CI f 2.17 to NA ). Thirty patients developed non-BM without any evidence of BM until time of death or last follow-up. Competing risk analysis identified 3 genes that were downregulated differentially in the patients with BM vs non-brain metastatic disease: CD37 (0.0170), cystatin A (0.0221), and IL-23A (0.0271). All 3 identified genes have plausible mechanisms for increasing metastatic phenotype. Other factors strongly associated with BM include: stage T (p = < 8.29e-6),stage N (p= 6.8e-4); adenocarcinoma vs squamous did not have a enough power to detect a difference (p =0.25). We have identified 3 genes CD37, cystatin A, and IL-23A for which downregulation led to a greater propensity for developing BM. Validation of these biomarkers could have implications on surveillance patterns of the brain.