Construction And Validation Of A Prognostic Model Of Nomogram In Metastatic Paraganglioma Patients Based On The SEER Database

Objective:Metastatic paraganglioma(PGL)is a rare neuroendocrine tumor of neural crest origin,and few tools are available to predict the prognosis of these patients.The objective of this study was to investigate the independent prognostic factors affecting patients with metastatic paraganglioma,and to develop and validate a nomogram clinical prediction model for metastatic PGL to accurately determine the prognosis of patient survival,for clinical purposes.Materials and methods:This study was conducted from the Surveillance,Epidemiology,and End Results(SEER)database which included 378 patients with metastatic PGL as the training group.A total of 79 patients with diagnosed PGL from January 2009 to December2021 were also collected from the First Affiliated Hospital of Nanchang University,13 of which had definite distant metastases as the external validation group.First,population differences in metastatic PGL were analyzed by comparing the demographics of the two cohorts using a chi-square test.The training group was then subjected to a one-way survival analysis with the long-rank method to test for differences between groups,followed by a multifactorial COX regression analysis to screen for independent risk factors for metastatic PGL survival.Lastly,a clinical prognostic prediction model for metastatic PGL was developed by R software,and the differentiation,accuracy and net clinical benefit of this Nomogram plot were evaluated by C-index,calibration curve and clinical decision analysis curve(DCA).Results:A total of 378 patients in the training set and 13 patients in the validation set were included.There were 173 males and 205 females in the taining group with a medium age of 46.5 years.Among the patients in the validation set,the medium age was 47 years,with 9 males and 4 females.The results of the survival analysis were combined by univariate and multifactorial Cox regression analysis,and the final inclusion of age(OR=1.80,95% CI:1.29-2.52),primary site(OR=2.08,95% CI:1.36-3.20),tumor stage(OR=1.763,95% CI:1.14-2.74),surgery(OR=1.46,95% CI:0.49-0.96),and chemotherapy(OR=1.87,95% CI:1.28-2.76)were independent predictors(P<0.05).The prognostic prediction model was constructed from the above five factors,and the results were presented as column line plots.The C-index(95% CI)of this prediction model was 0.718(0.673-0.748),and the model performed well in terms of discrimination.The corresponding calibration curves and DCA indicated that it was highly accurate and had some net clinical benefit.Conclusion:In this study,we established a clinical prediction model to predict the survival of patients with metastatic PGL through R language,which can provide clinicians with relatively accurate 1-year,3-year and 5-year OS predictions and serve as a reference for clinicians to evaluate the prognosis of patients with metastatic PGL.