Risk assessment model for first-cycle chemotherapy-induced neutropenia in patients with solid tumours

Abstract

Chemotherapy-induced neutropenia, the major dose-limiting toxicity of chemotherapy, is directly associated with concomitant morbidity, mortality and health-care costs. The use of prophylactic granulocyte colony-stimulating factors may reduce the incidence and duration of chemotherapy-induced neutropenia, and is recommended in high-risk patients. The objective of this study was to develop a model to predict first-cycle chemotherapy-induced neutropenia (defined as neutropenia grade>or=3, with or without body temperature>or=38 degrees C) in patients with solid tumours. A total of 1194 patients [56% women; mean age 58+/-12 years; 94% Eastern Cooperative Oncology Group (ECOG) status<or=1] with solid tumours were included in a multi-centre non-interventional prospective cohort study. A predictive logistic regression model was developed. Several factors were found to influence chemotherapy-induced neutropenia. Higher ECOG status values increased toxicity (ECOG 2 vs. 0, P=0.003; odds ratio 3.12), whereas baseline lymphocyte (P=0.011; odds ratio 0.67) and neutrophil counts (P=0.026; odds ratio 0.90) were inversely related to neutropenia occurrence. Sex and treatment intention also significantly influenced chemotherapy-induced neutropenia (P=0.012). The sensitivity and specificity of the model were 63% and 67% respectively, and the positive and negative predictive values were 17% and 94% respectively. Once validated, this model should be a useful tool for clinical decision making.

Figure 2

Logistic regression model equation for the prediction of first-cycle chemotherapy-induced neutropenia. Sex (1), male; Lym, baseline lymphocyte count; TI (1), treatment intention: radical adjuvant; TI (2), treatment intention: radical neo-adjuvant; TI (3), treatment intention: radical curative; ECOG (1), 1; ECOG (2), 2; Neutrop, baseline neutrophil count. ECOG, Eastern Cooperative Oncology Group.

Figure 1

Determining model sensitivity and specificity with the receiver operating characteristic curve (ROC).