Purpose: Febrile grade four (ie, < or = 500/microL) neutropenia (FN) is a frequent life-threatening complication of cancer chemotherapy. Although its incidence correlates to the dose of chemotherapy, FN may occur after almost any cytotoxic regimen. At present, there is no predictive method to identify patients who will experience FN.
Patients and methods: Univariate and multivariate analyses of risk factors for FN were performed on a retrospective cohort of 112 consecutive patients treated with various chemotherapy regimens. Two independent risk factors were identified by the logistic regression and used to create a risk model for FN. The validity of the model was tested in three distinct groups of patients: two prospective groups of patients treated in two institutions (Centre Léon Berard [CLB] and Institut G. Roussy [IGR]) and the group of patients with intermediate- or high-grade non-Hodgkin's lymphoma (NHL) treated with the doxorubicin, cyclophosphamide, vindesine, bleomycin, and prednisone (ACVBP) regimen between 1988 and 1992 at CLB.
Results: Within the retrospective group, 23 of 47 (49%) patients with lymphocyte counts < or = 700/microL at day 5 after chemotherapy experienced FN compared with seven of 65 (11%) of other patients (P = .00002). The type of chemotherapy (high dose v others) was also significantly correlated to FN (48% v 11%, P = .0003). Age, performance status, the number of previous chemotherapy cycles, or polymorphonuclear leukocyte (PMN) counts, were not significantly correlated to the incidence of FN in univariate analyses. Two independent risk factors were identified in the logistic regression: day 5 lymphocyte counts (beta = 1.97 +/- 0.53) and the type of chemotherapy regimen (beta = 1.91 +/- 0.53). The calculated probability to experience FN in patients with none, one, and both of these risk factors was 4.3%, 24.0%, and 68.8%, respectively. The validity of this model was tested in the three groups of patients used as validation samples. The observed incidences of FN in the above defined risk subgroups were 3%, 19%, and 67%, respectively, within the CLB prospective series and 6%, 19%, and 75% within the IGR prospective series. In the ACVBP group, the incidence of FN was 33% and 72%, respectively, in patients from the intermediate- and high-risk groups. In the two prospective groups and in the ACVBP series, the observed numbers of FN in the different risk groups did not differ significantly from those calculated by the model (P = .89, P = .86, and P = .72 for these three groups, respectively).
Conclusion: Day 5 lymphocyte counts < or = 700/microL and the type of chemotherapy regimen enable oncologists to define subgroups of patients treated with chemotherapy as those with a high intermediate, and low risk of FN. These criteria could be used to select subjects in whom prophylactic measures for FN, in particular hematopoietic growth factors, should be proposed.