Effectiveness and external validation of a prognostic prediction model for diffuse large B-cell lymphoma

To analyze the related factors affecting the prognosis of patients with diffuse large B-cell lymphoma (DLBCL), and to establish a Nomogram for prediction of the prognosis of patients with DLBCL.

The baseline data of 9 486 patients with DLBCL were extracted from the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute, and the SEER data were randomly divided into SEER training set (n=6 681) and SEER validation set (n=2 805) according to the ratio of 7 to 3. Univariate and multivariate Cox regression analyses were used to determine the risk factors affecting the prognosis of patients with DLBCL, and on this basis, a Nomogram was established; the discrimination and prediction efficiency of the model were evaluated by consistency index (C-index), area under the curve (AUC) of receiver operating characteristic (ROC) curve and calibration curve. A total of 120 patients with DLBCL in the Department of Hematology and Department of Oncology in authors' hospital from January 2013 to December 2017 were selected as external validation set, and the external validation was performed to verify the Nomogram and analyze its influencing factors.

Multivariate Cox regression analysis showed that age ≥ 60 years old, male, high stages of Ann Arbor staging, presence of B symptoms, lack of radiotherapy and lack of chemotherapy were the risk factors affecting the prognosis of DLBCL (P < 0.05), and the above factors were included in the Nomogram model. The C-index values of SEER training set, validation set and external validation set were 0.681, 0.669 and 0.817 respectively, and the AUC values of ROC curves of SEER training set, validation set and external validation set for predicting 5-year survival rate of DLBCL patients were 0.699, 0.678 and 0.869 respectively. The calibration curve showed that the result of the Nomogram prediction model was in good consistency with the actual result. Patients were divided into low-risk group (< 170 points) and high-risk group (≥ 170 points) according to the calculated risk scores, and the 5-year survival rate of the low-risk group was the highest in the SEER training set, validation set and external validation set.

Male, age ≥ 60 years old, high stages of Ann Arbor staging, lack of radiotherapy, lack of chemotherapy and presence of B symptoms are the independent risk factors affecting the prognosis of patients with DLBCL. The Nomogram prediction model established based on the above factors has good predictive performance, and can be used for personalized risk assessment and prediction of the prognosis of patients with DLBCL.