Prognostic nomogram for primary splenic lymphoma: a SEER database-based study.

This study aimed to establish a nomogram model based on the clinicopathological factors affecting the prognosis of patients with primary splenic lymphoma (PSL) to predict the overall survival (OS) and cancer-specific survival (CSS) of patients. A total of 4074 patients diagnosed with PSL were included in this study. Among them, 4052 cases from the SEER (Surveillance, Epidemiology, and End Results) database were randomized into a training set and an internal validation set in a 7:3 ratio. Another 22 patients from the First Affiliated Hospital of Xi'an Jiaotong University were used as an external validation set. The prognostic factors affecting the OS and CSS of patients were analyzed using univariate and multivariate Cox regression models. Survival analysis was performed using Kaplan-Meier (KM) method and compared by Log-rank test. Then, a nomogram model was established to predict OS and CSS. Finally, the model was validated both internally and externally using the concordance index (C-index), receiver operating characteristic curve (ROC), and calibration curve to evaluate its predictive value, and the decision curve analysis (DCA) was conducted to assess its clinical utility. Our results showed that the model displayed a good prediction ability. In the training set, the OS rates at 1, 3, and 5 years were 85.9%, 75.8% and 70.1%, respectively, while the CSS rates at 1, 3, and 5 years were 91.9%, 86.2% and 82.3%, respectively. Predictors in the prediction model of OS included age, sex, marital status, Ann Arbor stage, histology, surgery, chemotherapy and year at diagnosis. On the other hand, predictors in the model of CSS included age, Ann Arbor stage, histology, chemotherapy, and year at diagnosis. Internal and external validation of the nomogram model showed that the C-index for predicting OS was 0.678 (0.662, 0.694) in the training set, 0.672 (0.648, 0.696) in the internal validation set, and 0.704 (0.565, 0.843) in the external validation set; the C-index for predicting CSS was 0.685 (0.661, 0.709) in the training set, 0.683 (0.650, 0.716) in the internal validation set, and 0.676 (0.488, 0.864) in the external validation set. The calibration curves for several groups showed good consistency, and DCA suggested its clinical usability. In conclusion, the nomogram constructed in this study has a good predictive value for the survival of patients with PSL, and can be a clinically applicable and practical prediction tool, facilitating rapid and accurate individualized predictions of the patient survival.