A nomogram based on the expression level of angiopoietin-like 4 to predict the severity of community-acquired pneumonia

BACKGROUND: The morbidity and mortality of community-acquired pneumonia (CAP) remain high among infectious diseases. It was reported that angiopoietin-like 4 (ANGPTL4) could be a diagnostic biomarker and a therapeutic target for pneumonia. This study aimed to develop a more objective, specific, accurate, and individualized scoring system to predict the severity of CAP. METHODS: Totally, 31 non-severe community-acquired pneumonia (nsCAP) patients and 14 severe community-acquired pneumonia (sCAP) patients were enrolled in this study. The CURB-65 and pneumonia severity index (PSI) scores were calculated from the clinical data. Serum ANGPTL4 level was measured by enzyme-linked immunosorbent assay (ELISA). After screening factors by univariate analysis and receiver operating characteristic (ROC) curve analysis, multivariate logistic regression analysis of ANGPTL4 expression level and other risk factors was performed, and a nomogram was developed to predict the severity of CAP. This nomogram was further internally validated by bootstrap resampling with 1000 replications through the area under the ROC curve (AUC), the calibration curve, and the decision curve analysis (DCA). Finally, the prediction performance of the new nomogram model, CURB-65 score, and PSI score was compared by AUC, net reclassification index (NRI), and integrated discrimination improvement (IDI). RESULTS: A nomogram for predicting the severity of CAP was developed using three factors (C-reactive protein (CRP), procalcitonin (PCT), and ANGPTL4). According to the internal validation, the nomogram showed a great discrimination capability with an AUC of 0.910. The Hosmer–Lemeshow test and the approximately fitting calibration curve suggested a satisfactory accuracy of prediction. The results of DCA exhibited a great net benefit. The AUC values of CURB-65 score, PSI score, and the new prediction model were 0.857, 0.912, and 0.940, respectively. NRI comparing the new model with CURB-65 score was found to be statistically significant (NRI = 0.834, P < 0.05). CONCLUSION: A robust model for predicting the severity of CAP was developed based on the serum ANGPTL4 level. This may provide new insights into accurate assessment of the severity of CAP and its targeted therapy, particularly in the early-stage of the disease.