Logistic regression analysis for febrile neutropenia (FN) induced by docetaxel in Japanese cancer patients

Abstract

Fever occurring in a neutropenic patient remains a common life-threatening complication of cancer chemotherapy, and febrile neutropenia (FN) is recognized as a dose-limiting factor (DLF) in cancer chemotherapy. The aim of this study is to evaluate the significant covariate associated with the risk of FN occurrence in Japanese patients. A stepwise logistic regression was conducted using data from Japanese cancer patients treated with docetaxel. Based on those results, an equation was established which predicts the probability of FN occurrence. From the result of a stepwise multivariate logistic regression analysis, performance status factor (PS*), which is set to 1 if performance status factor is 2 or 3, and to 0 otherwise and area under the plasma concentration versus time curve (AUC) were selected as covariates significantly associated (p < 0.05) with FN occurrence. The obtained equation to predict the probability (P) of docetaxel-induced FN occurrence is P = 1/[1 + exp{-(1.29 x AUC + 1.41 x PS* -3.52)}]. A receiver operating characteristic (ROC) curve analysis revealed that the best cut-off value of FN probability to differentiate between the presence and absence of FN was 0.61. An equation was developed to predict the probability of FN occurrence for Japanese patients treated with docetaxel. It was found that FN may not occur when the probability of FN occurrence calculated by the predictive equation is less than 0.61. Therefore, the predictive equation for FN occurrence may be used for selecting the appropriate dose to avoid the occurrence of FN.