Construction of a prognostic model for new diffuse large B-cell lymphoma and research on risk factors for secondary central nervous system infiltration

Abstract Objective: This study aims to improve the International Prognostic Index (IPI) and enhance the Central Nervous System-International Prognostic Index (CNS-IPI) for patients with diffuse large B-cell lymphoma (DLBCL) in the northern Chinese population. Methods: A total of 1,364 DLBCL patients were collected from January 2000 to December 2019 at the Department of Lymphoma of Peking University Cancer Hospital in Beijing and the Department of Lymphoid Oncology at the Third Hospital of Shanxi Medical University in Shanxi Province, China, for analysis, in addition to the five factors from the IPI—age, stage, performance status (ECOG), lactate dehydrogenase (LDH), and extranodal involvement (E) number—fourteen other extranodal sites and progression within six months (POD6) were included, for a total of 20 factors in a nomogram prognostic model for overall survival (OS) in DLBCL patients. The model's stability at different time points was verified using receiver operating characteristic (ROC) curves, calibration curves, the time-dependent concordance index (C-index), and the time-dependent area under the ROC curve (AUC). Expression levels of MYC, BCL2, P53, BCL6 protein, and the co-expression of MYC and BCL2 (Double Expressed) were detected by immunohistochemistry (IHC); Double hits (DHT) were identified via FISH. MYC, BCL2, P53, BCL6, DHT, and next-generation sequencing (NGS) results were subjected to univariate and multivariate regression analyses to assess their impact on prognosis. A high-risk factor analysis was conducted among 54 patients with secondary central nervous system (SCNS) lymphoma who had concurrent CNS involvement. Results: Twenty clinical characteristics were used to build a nomogram model predicting OS in DLBCL patients. The nomogram indicates that POD6 is the main high-risk factor for DLBCL (HR = 5.13, 95% CI 1.03-2.51, p < 0.0001). IPI factors—Age, Stage, LDH, and Extra-nodal invasion—remain poor prognostic indicators aside from ECOG, with extra-CNS invasion being the most detrimental of all 14 extra-nodal lesions (HR=1.61, 95% CI 1.03-2.51, p=0.035). The model achieved AUC values over 0.7 at 1, 3, and 5 years. Calibration curves showed a consistent distribution of points on both sides of the diagonal line, with slopes close to 1, which is statistically significant, indicating high predictive accuracy. Among protein expressions and gene mutations, TP53 and MYC mutations are key adverse factors. POD6 in CNS involvement was added to the CNS-IPI. Conclusion: Compared to the IPI, this study developed a new prognostic model for DLBCL patients, incorporating POD6 as a novel high-risk factor. The model significantly enhances the accuracy of clinical prognostic prediction. POD6 also allows for predicting central nervous system involvement. TP53 and MYC gene mutations are key adverse factors. The model is clinically meaningful and has broad applicability in practice. Keywords: Diffuse Large B Cell Lymphoma (DLBCL), Nomogram, Prognostic models, POD6, TP53 and MYC gene mutations