To identify key genes associated with tumor-associated macrophages (TAMs), tumor immunotherapy, in the prognosis of lung adenocarcinoma (LUAD). The mRNA expression profiles of LUAD samples were obtained from The Cancer Genome Atlas (TCGA) database. The "CIBERSORT" R package was employed to calculate the proportion of innate immune cell infiltration in both tumor and adjacent normal tissues. TAM-associated genes in LUAD were identified to construct a prognostic risk model using weighted gene correlation network analysis (WGCNA), Least Absolute Shrinkage and Selection Operator (LASSO), and multivariate Cox regression analyses (COX). The IMvigor210 cohort was utilized to validate the roles of these genes as predictors of immunotherapy response. Tissue microarrays, immunofluorescence staining, and mRNA level detection methods were used to determine the correlation of risk factors in LUAD tissues. CIBERSORT analysis revealed significant differences in innate immune cells between tumor and adjacent tissues. Seventy-four differential genes linked to these cells were identified from WGCNA. Four hub genes (endothelin receptor type B, vascular endothelial growth factor D (VEGFD), latent transforming growth factor beta binding protein 4 (LTBP4), and fibroblast growth factor receptor 4 (FGFR4)) in the TAM prognostic model were identified as independent prognostic risk factors (P < 0.05). VEGFD expression was identified as a low-risk factor for LUAD prognosis prediction (P < 0.05). Moreover, low-risk patients exhibited higher sensitivity to anti-PD-L1 therapy compared to high-risk patients (P < 0.05). VEGFD levels were negatively correlated with programmed cell death 1 (PD-1) levels (r = -0.363; P < 0.05), suggesting that VEGFD may serve as a predictor for anti-PD-1 treatment. VEGFD is associated with innate immunity in LUAD, it can predict LUAD prognosis, and therefor may be a potential predictor for anti-PD-1 treatment in patients with LUAD.