Computational recognition of lncRNA signature of tumor-infiltrating activated CD4+ memory T cells with potential implications in prognosis and immunotherapy of lung adenocarcinoma.

e20519 Background: CD4 memory T cells are an important components of the tumour microenvironment and influence tumourigenesis and progression. However, it has not been determined whether lncRNAs in lung adenocarcinoma (LUAD) play a role in activated CD4+ memory T cells (CD4amT). Methods: The LUAD dataset acquired by The Cancer Genome Atlas was used as a training cohort (TCGA-LUAD). For the validation cohort, we used the combination of lung adenocarcinoma datasets from GENE EXPRESSION OMNIBUS database acquired 2 LUAD datasets, and set as the validation cohort (GEO-LUAD). Lung cancer immunotherapy cohorts have been published to test immune responses. The CIBERSORT algorithm was used to assess immune cell infiltration of tumours. Univariate Cox analysis was used to screen for prognostic CD4amT-related lncRNAs (CD4amTRL). LASSO analysis and multivariate Cox analysis were then performed to construct model for prognostic lncRNAs, the effects of which were assessed by Kaplan-Meier curves and subject operating characteristics (ROC). Patients were divided into high and low risk groups based on CD4amTRL signature. We have analyzed differences in enrichment pathways between the two groups and subsequently we investigated genes associated with the cancer-immune process, the immune microenvironment, immunotherapy responses and multiple anti-cancer drug sensitivity correlations. Results: In the two LUAD cohorts, compared with the high infiltration of CD4amT, those in the low infiltration group had better OS benefits (TCGA-LUAD: HR = 0.678; P = 0.028; GEO-LUAD: HR = 0.435; P < 0.0001). We identified 6 prognostic CD4amTRL and constructed prediction models. cd4 CD4amTRL signature distinguished significantly different OS between the high-risk and low-risk groups of TCGA-LUAD (HR = 4.26). The AUCs predicting 1-, 3- and 5-year survival were 0.737, 0.715 and 0.722, respectively. In the GEO-LUAD dataset, the HR was 2.03, while the AUCs predicting 1-, 3- and 5-year survival were 0.719, 0.681 and 0.630, respectively. The risk score model independently predicted OS in LUAD patients and was associated with immune checkpoints. Numerous genes related to the cancer-immune process were highly correlated. In addition, the risk model was significantly associated with multiple immune microenvironment features (p < 0.001). OS was significantly different between the two immunotherapy cohorts in the high- and low-risk groups for CD4amTRL signature (P < 0.001). Conclusions: Our novel risk model developed based on CD4amTRL signatures can aid in predicting clinical prognosis and guiding treatment in patients with LUAD.