Identification and Validation of a Prognostic Prediction Model in Diffuse Large B-Cell Lymphoma

BackgroundDiffuse large B-cell lymphoma (DLBCL) is a heterogeneous group with varied pathophysiological, genetic, and clinical features, accounting for approximately one-third of all lymphoma cases worldwide. Notwithstanding that unprecedented scientific progress has been achieved over the years, the survival of DLBCL patients remains low, emphasizing the need to develop novel prognostic biomarkers for early risk stratification and treatment optimization. MethodIn this study, we screened genes related to the overall survival (OS) of DLBCL patients in datasets GSE117556, GSE10846, and GSE31312 using univariate Cox analysis. Survival-related genes among the three datasets were screened according to the criteria: hazard ratio (HR) >1 or p-value <0.01. Least Absolute Shrinkage and Selection Operator (LASSO) and multivariate Cox regression analysis were used to optimize and establish the final gene risk prediction model. The TCGA-NCICCR datasets and our clinical cohort were used to validate the performance of the prediction model. CIBERSORT and ssGSEA algorithms were used to estimate immune scores in the high- and low-risk groups. ResultsWe constructed an eight-gene prognostic signature that could reliably predict the clinical outcome in training, testing, and validation cohorts. Our prognostic signature also performed distinguished areas under the ROC curve in each dataset, respectively. After stratification based on clinical characteristics such as cell-of-origin (COO), age, eastern cooperative oncology group (ECOG) performance status, international prognostic index (IPI), stage, and MYC/BCL2 expression, the difference in OS between the high- and low-risk groups was statistically significant. Next, univariate and multivariate analyses revealed that the risk score model had a significant prediction value. Finally, a nomogram was established to visualize the prediction model. Of note, we found that the low-risk group was enriched with immune cells. ConclusionIn summary, we identified an eight-gene prognostic prediction model that can effectively predict survival outcomes of patients with DLBCL and built a nomogram to visualize the perdition model. We also explored immune alterations between high- and low-risk groups.

Automated summary

In this study, we established a prognostic prediction model with eight genes that can effectively predict the survival outcomes of DLBCL patients. We also created a nomogram to visualize the prediction model and explored immune alterations between high- and low-risk groups.