Predictors of Febrile Neutropenia Among Cancer Patients Receiving Myelosuppressive Chemotherapy in US Clinical Practice

Introduction: Febrile neutropenia (FN) is a common and serious complication of myelosuppressive chemotherapy. Assessment of FN risk has typically focused on the chemotherapy course or the first cycle of chemotherapy, and risk factors present at the time of chemotherapy initiation. However, a substantial proportion of FN episodes occur after the first chemotherapy cycle, and it is likely that the risk of these episodes is determined-at least in part-by events/care occurring during the course of chemotherapy (e.g., prior FN, chemotherapy dose reductions, colony-stimulating factor [CSF] prophylaxis). The objectives of this study were two-fold: (1) to evaluate FN risk factors that are ascertained at chemotherapy initiation as well as those ascertained on a cycle-specific basis during the chemotherapy course; and (2) to evaluate how use of CSF prophylaxis may reduce this risk.

Methods: A retrospective cohort design and data from four US health systems-Henry Ford Health System, Kaiser Permanente Northwest, Reliant Medical Group, and Geisinger Health System-were employed. Data included patients' demographics and clinical profile, cancer, treatment, and outcomes, and were compiled from electronic medical records systems, administrative data warehouses, medical charts, and cancer registries. The study population comprised all patients who received a course of myelosuppressive chemotherapy for breast cancer, colorectal cancer, lung cancer, or non-Hodgkin's lymphoma (NHL) from 2009-2017. For each patient, each cycle of chemotherapy during the course was characterized and included as a separate observation in the analytic file; thus, each patient could contribute multiple chemotherapy cycles to the analysis. Candidate predictors were identified from national guidelines and published literature; cycle number, FN events occurring prior to the cycle of interest, chemotherapy relative dose intensity (RDI), and CSF prophylaxis were defined on a cycle-specific basis. Independent risk factors for FN were evaluated within a multivariable framework using generalized estimating equations with a logistic link function.

Results: The study population included 4091 patients who received myelosuppressive chemotherapy for breast cancer (49%), colorectal cancer (17%), lung cancer (23%), or NHL (11%), and who contributed 29,964 cycles of observation; mean age was 61 years, 36% had metastatic disease, 36% received CSF prophylaxis in cycle 1, and 46% received prophylaxis in ≥1 cycle. During the course, 13.8% of patients developed FN; risk was greatest in cycle 1 (5.2%). Independent predictors of FN included patient, cancer, and chemotherapy characteristics ascertained at the time of chemotherapy initiation as well as those ascertained on a cycle-specific basis during the course; in particular, cycle-specific FN odds were markedly higher for regimens with an intermediate/high/unclassified FN risk level (odds ratio [OR] = 1.5-1.8), higher chemotherapy RDI (OR = 2.0), and prior FN (OR = 3.7), and were lower with use of CSF prophylaxis (OR = 0.60) (Table).

Conclusions: In this retrospective evaluation of patients receiving myelosuppressive chemotherapy for breast cancer, colorectal cancer, lung cancer, or NHL, a number of risk factors were found to be important in predicting FN at the cycle level, most notably those that were ascertained on a cycle-specific basis during the chemotherapy course (e.g., chemotherapy RDI, prior FN). In addition, FN odds were found to be substantially higher for chemotherapy regimens with a high, intermediate, or unclassified (vs. low) FN risk level, and were found to be lower among patients who received CSF prophylaxis in that cycle. Careful consideration should be given to identifying patients at elevated risk of FN before administration of chemotherapy in all cycles.

View largeDownload slide

View largeDownload slide

Close modal