Identification and Verification of Immune-Related Gene Prognostic Signature Based on Ssgsea for DLBCL

Background : Diffuse large B-cell lymphoma (DLBCL) is the most common form of non-Hodgkin's lymphoma (NHL), accounting for about 30% of NHL worldwide and a higher incidence of about 50% in China. The tumor microenvironment (TME) and effective immune surveillance are important in the pathogenesis of lymphoma and are closely related to the prognosis.

Patients and methods: In this study, we evaluated the immune cell infiltration of 223 patients with diffuse large B-cell lymphoma (DLBCL) downloaded from GEO using ssGSEA and evaluated the RNA sequencing of these patients, divided the GEO downloaded dataset into high and low immune groups, performed differential gene analysis.

Results: 952 genes were found to be upregulated. Then we grouped our patients' sequencing data into high and low immune groups. The same differential gene analysis was performed, and upregulation of 425 genes was found. We discovered 22 duplicated genes among these genes yVenn diagram. and the expression of these 22 genes was differentially expressed in the high and low immune groups. 4 genes associated with poor patient prognosis were identified by one-way COX and LASSO regression (CELF2, KIFC1, MEGF6, S100A9), especially KIFC1. We used such four genes as prognostic features to construct the model, and chose GSE10846 as the Train group and GSE87371 as the Test group to verify the accuracy of the model. In the Train and Test groups, there was a difference in survival between high and low risk groups (p<0.05), indicating that the ROC curves suggest that the model can accurately predict patient survival. The 1-year, 2-year, and 3-year calibration curves also show that our model can accurately predict the survival index our patients. The risk curves and survival status plots indicate that the number of death increases as the risk score increases, in summary indicating that our model is able to accurately predict the prognosis of the disease. Next, we analyzed KIFC1 and found that the expression of KIFC1 differed between high and low immune groups. Correlation analysis showed that KIFC1 was positively correlated with patient risk scores. In the GSE32018 dataset, the expression of KIFC1 in tumor tissues differed significantly from normal tissues, and patients in the high expression group was less likely to survive. Univariate and multivariate analyses showed that KIFC1 could be used as an independent prognostic factor. Clinical correlation analysis showed that KIFC1 correlated with patients' extra-nodal involvement, ECOG grade, and tumor subtype.

Conclusion: The results of provide a novel approach to predict the prognosis and survival of DLBCL patients by these genetic features. Also, KIFC1 could potentially be used as a therapeutic target.

No relevant conflicts of interest to declare.