Risk of Febrile Neutropenia Associated With Select Myelosuppressive Chemotherapy Regimens in a Large Community-Based Oncology Practice.

NCCN has classified commonly used chemotherapy regimens into high (>20%), intermediate (10%-20%), or low (<10%) febrile neutropenia (FN) risk categories based primarily on clinical trial evidence. Many chemotherapy regimens, however, remain unclassified by NCCN or lack FN incidence data in real-world clinical practice. We evaluated incidence proportions of FN and grade 4 and 3/4 neutropenia during the first chemotherapy course among patients from Kaiser Permanente Southern California who received selected chemotherapy regimens without well-established FN risk. Patients given granulocyte colony-stimulating factor (G-CSF) prophylaxis were excluded. Sensitivity analyses were performed to account for FN misclassification and censoring. From 2008 to 2013, 1,312 patients with breast cancer who received docetaxel and cyclophosphamide (TC; n=853) or docetaxel, carboplatin, and trastuzumab (TCH; n=459); 1,321 patients with colorectal cancer who received capecitabine and oxaliplatin (XELOX; n=401) or leucovorin, 5-fluorouracil, and oxaliplatin (FOLFOX6; n=920); 307 patients with non-Hodgkin's lymphoma who received bendamustine with or without rituximab; and 181 patients with multiple myeloma who received lenalidomide with or without dexamethasone were included. Crude FN risk was >20% for both breast cancer regimens (TC and TCH). Crude FN risks for XELOX, FOLFOX6, bendamustine, and lenalidomide were <10%; however, when potential FN misclassification and censoring were considered, FN risks were >10%. Our results support published literature highlighting the real-world, "high" FN risk of the TC and TCH regimens for breast cancer. There is strong suggestive evidence that FN risks for XELOX, FOLFOX6, bendamustine, and lenalidomide are >10%. Calculation of chemotherapy course-level FN incidence without controlling for differential censoring for patients who discontinued regimens early, or possible FN misclassification, might have resulted in bias toward an underestimation of the true FN risk. These findings help define FN risk of the selected regimens in the real-world setting and inform prophylactic G-CSF use.