Immunotherapy outcomes and profile in lung cancer brain metastases.

8646 Background: Brain metastasis (BM) in NSCLCs have a unique tumor micro-environment (TME) characterized by distinct immune-constituents and the presence of blood-brain barrier, differentiating them from primary NSCLC tumors (PT). Despite the high incidence of BM, earlier trials indicated that less than 15% of BM patients benefit from immune checkpoint inhibitors (ICIs). We characterized the TME of BM to potentially identify a subset of patients who may respond favorably to ICIs. Methods: 36,726 NSCLC samples were analyzed by NGS of DNA (592-gene) and RNA (whole transcriptome) at Caris Life Sciences (Phoenix, AZ). TMB-high was defined as >=10mt/MB, PD-L1 was tested by IHC (22c3), and TME by RNA expression (quanTIseq). Real-world overall survival was obtained from insurance claims data and calculated from treatment start on six common ICIs to last contact (IO-OS), while time-on-treatment (TOT) was from first to last of treatment. Hazard ratios (HRs) were calculated using the Cox proportional hazards model and p values (log-rank test). Fisher’s exact tests and Mann-Whiney U were used and adjusted for multiple comparisons (q<0.05). Results: Samples were comprised of 63.5%, PT 4.9% BM, and 31.6% extracranial metastases. High TMB and loss of heterozygosity (LOH) rates were higher in BM vs. PT (55% vs. 35.7%) and (28.0% vs. 14.3%), q<0.05, respectively, while PD-L1+ rate was similar (53.3% BM vs. 55.6% PT). Significant mutational differences in BM vs. PT included TP53 (77.7% vs 65.1%), KEAP1 (19.4% vs. 12.1%), and STK11 (17.5% vs. 12.0%), q<0.05. Immune deconvolution demonstrated the abundance of neutrophils, Tregs, monocytes, and B cells was higher in PT vs. BM (fold-change: 1.35-1.73, q<0.05), while dendritic cells were higher in BM vs. PT (fold-change: 13, q<0.05). BM had significantly longer IO-OS (24.4 vs. 19.7 mo: HR = 0.83, 95% CI 0.76-0.90, p<0.05) and TOT (7.7 vs. 6.0 mo; HR = 0.83, 95% CI 0.77-0.89, p<0.05) compared to PT. PD-L1+ and TMB High were associated with longer IO-OS in BM vs. PT (HR = 0.71, 95% CI 0.63-0.81, p<0.05) and (HR = 0.83, 95% CI 0.74 – 0.93, p<0.05), respectively. By histology, squamous cell carcinoma (SC) was significantly more prevalent in PT compared to BM (25.9% vs. 8.5%, q<0.05). In adenocarcinoma (AD), BM was associated with longer IO-OS (HR = 0.78, 95% CI 0.70-0.87, p<0.05) compared to PT, while in SC, the survival association was not significant. The longer survival seen in AD was still significant after multivariate analysis with TP53, STK11 and KEAP1 (HR = 0.76, 95% CI 0.68-0.86, q<0.05). In BM, PD-L1+ and TMB High were associated with longer IO-OS in AD and may be considered first line treatment in this population, if asymptomatic. Conclusions: Historically BM have been associated with poor outcomes. We demonstrate that BM has higher TMB, increased infiltration of dendritic cells, and higher LOH than PT. After stratifying patients based on PD-L1 and TMB, patients with BM adenocarcinoma histology had improved post ICI survival.