Comprehensive analysis of genomic alterations and novel prognostic biomarkers, and establishment of prediction models of metastasis in metastatic non-small cell lung cancer

<b>Introduction:</b> Most patients with non-small cell lung cancer (NSCLC) have metastases at initial diagnosis. However, the comprehensive molecular characteristics and factors associated with its metastases are still needed. <b>Methods:</b> Tumor sequencing of 556 cancer-related genes was performed on 114 Chinese NSCLC patients. A distinct genomic profile was identified in metastatic patients compared to those without metastases. Kaplan-Meier method was used to analyze the associations between clinical outcomes, clinical characteristics, and mutated genes. The Fisher test and Lasso logistic regression analysis were employed to identify factors related to metastasis and to develop prediction models. <b>Results:</b> Male, squamous cell lung carcinoma, and smokers showed strikingly higher TMB levels in all NSCLCs. The metastatic group had a significantly higher proportion of patients aged ≥ 70 years and in stage III-IV. <i>TP53</i> was the most frequent mutation in both groups, and <i>EGFR</i> tended to be higher in the metastatic group. The copy number variation events occurred more frequently in the metastatic group. Additionally, predictive models for metastasis (AUC = 0.828), pleural metastasis (AUC = 0.582), and multisite metastasis (AUC = 0.559) were established. Females, and <i>EGFR</i> +, <i>ASXL2-</i>, and <i>STK11-</i> cases had better overall survival (OS). Lung adenocarcinoma, and <i>KMT2D-</i> and <i>STK11-</i> cases had better progression-free survival (PFS). NSCLC metastasis was associated with poor OS and poor PFS. <b>Conclusions:</b> Our study provided a comprehensive analysis of genomic alterations in metastatic NSCLCs, identified novel prognostic biomarkers, and provided three predictive models for metastasis, which may have potential implications for personalized treatment strategies.