An immune scoring system predicts prognosis and immune characteristics in lung adenocarcinoma brain metastases by RNA sequencing

Previous studies have reported that the tumor immune microenvironment (TIME) was associated with the prognosis of lung cancer patients and the efficacy of immunotherapy. However, given the significant challenges in obtaining specimens of brain metastases (BrMs), few studies explored the correlation between the TIME and the prognosis in patients with BrMs from lung adenocarcinoma (LUAD).

We identified two immune-related subtypes which used to estimate prognosis and immune characteristics and developed a reliable GYMS-related risk signature in LUAD BrMs. These results will enhance the understanding of the immune microenvironment in LUAD BrMs and lay the theoretical foundation for the development of personalized therapies for LUAD patients with BrMs.

Transcript profiling of archival formalin-fixed and paraffin-embedded specimens of BrMs from 70 LUAD patients with surgically resected BrMs was carried out using RNA sequencing. An immune scoring system, the green-yellow module score (GYMS), was developed to predict prognosis and immune characteristics in both BrMs and primary LUAD using Weighted Correlation Network analysis (WGCNA) and GSVA analysis. We comprehensively evaluated the immunological role of GYMS based on gene expression profile of LUAD BrMs by systematically correlating GYMS with immunological characteristics and immunotherapy responsiveness in the BrMs. Immunohistochemistry was applied for validation.

We found that the high-GYMS group had better clinical prognosis and inflamed immune landscape including high infiltrations of various immune cells, increased immunomodulatory expression, and enriched immune-related pathways by using RNA-seq and immunohistochemical analysis. Low-GYMS group presented a lacked immune infiltration characteristic. Besides, the high-GYMS group had lower TIDE score and higher T-cell inflamed score than low-GYMS group. The GYMS has been validated in independent BrMs cohorts and primary NSCLC cohort treated with anti-PD-1/PD-L1, showing strong reproducibility and stability in both primary LUAD and BrMs. In addition, we construct a GYMS-related risk signature for patients with LUAD BrMs to predict prognosis. Conclusions: We identified two immune-related subtypes which used to estimate prognosis and immune characteristics and developed a reliable GYMS-related risk signature in LUAD BrMs. These results will enhance the understanding of the immune microenvironment in LUAD BrMs and lay the theoretical foundation for the development of personalized therapies for LUAD patients with BrMs.