Predicting bacterial infection risk in patients with ANCA-associated vasculitis in southwest China: development of a new nomogram.

OBJECTIVES

The aim of this study was to develop and assess a risk nomogram of bacterial infection in patients with ANCA-associated vasculitis (AAV) in southwest China.

METHOD

We established a prediction model based on a training dataset of 249 AAV patients. The least absolute shrinkage and selection operator (Lasso) was used to screen feature variables. Multivariate logistic regression analysis was used to build a prediction model for feature variables. Nomogram was used to predict the risk of bacterial infection in AAV patients. Receiver operating characteristic (ROC) curve was used to evaluate and verify the prediction accuracy of the model. Calibration and clinical useful range was assessed using calibration curve and decision curve analysis, respectively.

RESULTS

Bactericidal permeability enhancement protein of ANCAs (BPI-ANCAs), procalcitonin (PCT), and white blood cell (WBC) were the characteristic variables in this study. Nomogram showed that positive BPI-ANCAs and PCT had higher positive predictive value for bacterial infection in AAV patients. The area under curve (AUC) of the model was 0.703 (95% confidence interval: 0.640-0.766). In the validation model, the AUC was 0.745 (95% confidence interval: 0.617-0.872). Decision curve analysis showed that the nonadherence nomogram was clinically useful within the threshold probability range of 0.31-0.85.

CONCLUSIONS

Nomogram combined with BPI-ANCAs and PCT has the guiding significance for predicting bacterial infection risk in AAV. As an ANCA-specific autoantibody, BPI-ANCAs is helpful for clinicians to understand the role of specific autoantibodies in the pathogenesis of AAV. Key Points • BPI-ANCAs, PCT, and WBC could predict bacterial infection in AAV patients. • Nomogram showed that positive BPI-ANCAs had a high positive predictive value for bacterial infection in AAV patients.