Establishment and assessment of a nomogram for predicting the differentiation syndrome of acute promyelocytic leukemia

Objective By analyzing the risk factors for occurrence of differentiation syndrome (DS) during induction therapy in newly-diagnosed acute promyelocytic leukemia (APL) patients, a prediction nomogram for DS was established and the accuracy of this nomogram was validated. Methods The modeling group was made up of 130 classical APL patients during the period of 1st January 2011 to 31st December 2013. After single factor screening of clinical variables, the logistic regression model was used to identify the final model variables. A nomogram subsequently established by R software was validated by Bootstrap resampling as internal validation. Concordance index (C-index) was used for the accuracy evaluation of the nomogram, and calibration curves were painted to test the actual observation and the nomogram-prediction of occurrence rate of DS. Results Occurrence rate of DS in 130 APL patients was 30.0%; In multivariate analysis, body mass index (BMI) ≥24 kg/m2 and without using steroids for prevention of DS were identified as independent risk factors. The C-index of the nomogram for predicting DS was 0.818 (95% CI 0.741-0.895). The calibration curves showed good concordance of occurrence rate of DS between nomogram-prediction and actual observation. Conclusion The nomogram was successfully established as a more accurate and visible tool for predicting the occurrence rate of DS in APL patients. Key words: Leukemia, promyelocytic, acute; Differentiation syndrome; Risk factors; Nomograms; Prediction model