Predicting the acute-phase response fever risk in bisphosphonate-naive osteoporotic patients receiving their first dose of zoledronate.

Introduction: To devise a precise and efficient tool for predicting the individualized risk of acute-phase response (APR) in bisphosphonate (BP)-naive osteoporotic (OP) patients, receiving their first intravenous dose of zoledronate (ZOL).

Methods: The baseline clinical and laboratory data of 475 consecutive BP-naive OP patients, who received their first intravenous dose of ZOL between March 2016 and March 2021 in the Affiliated Kunshan Hospital of Jiangsu University, were chosen for analysis. Univariate and multivariable logistic regression models were generated to establish candidate predictors of APR fever risk, using three distinct fever thresholds, namely, 37.3 °C (model A), 38.0 °C (model B), and 38.5 °C (model C). Next, using predictor regression coefficients, three fever-threshold nomograms were developed. Discrimination, calibration, and clinical usefulness of each predicting models were then assessed using the area under the curve (AUC), calibration curve (CC), and decision curve analysis (DCA). The internal and external model validations were then performed.

Results: The stable predictors were age, serum 25-hydroxy vitamin D, serum total calcium, and peripheral blood erythrocytes count. These were negatively associated with the APR fever risk. The AUCs of models A, B, and C were 0.828 (95% confidence intervals [CI], 0.782 to 0.874), 0.825 (95% CI, 0.767 to 0.883), and 0.879 (95% CI, 0.824 to 0.934), respectively. Good agreement was observed between the predictions and observations in the CCs of all three nomograms.

Conclusions: This study developed and validated nomogram prediction models that can predict APR fever risk in BP-naive OP patients receiving their first infusion of ZOL.