Frequency of Infection Source and Predictors of Mortality in Leukemia Patients Diagnosed with Febrile Neutropenia

Introduction:High-risk hematologic malignancies are associated with significant treatment-related morbidity, particularly due to infectious events. Neutropenic fever/Febrile neutropenia (NF) is one of the most common infectious complications of chemotherapy; however, mortality rates associated with NF and the factors predictive of mortality are not well-established in the adult leukemia literature. We sought to examine factors that influence mortality and length of stay in patients admitted with NF to the inpatient hematologic malignancy service of a tertiary care center.Materials and Methods:We performed a retrospective chart review of patients admitted to our hematologic malignancy service between April 2016 and July 2018 who carried a diagnosis of NF, based upon absolute neutrophil count < 500/microL and temperature =>38.3C. We included patients who developed fever during planned admission for treatment or who were admitted for complications. Patients had primary diagnoses of AML, ALL, CML, CLL, and MDS. Patients undergoing autologous or allogenic stem cell transplant were excluded. Our analysis included demographics, primary diagnosis, primary disease status (remission, relapse, refractory), line of therapy prior to admission, duration of neutropenia, and nutritional status. We included infection-specific factors such as the presence of bacteremia, identification of an infection source, gram stain of bacteria isolated, resistance patterns, and need for ICU care. We used linear regression analysis to evaluate potential predictors of length of stay and mortality. Statistical analyses were performed with SAS.Results:We identified 100 episodes of NF that occurred in 63 patients. Median age was 58. Malignancies with the highest representation among NF episodes were AML (69%), ALL (24%), CML (3%), and CLL (2%). Infection sources were identified in 62% of episodes, with bacteremia identified in 54% of cases, accounting for 87% of sources. Mean length of stay was 18.7 (+/- 2.3 days). 14 deaths occurred, with 13 (93%) being AML patients and 11 (79%) having bacteremia. Factors associated with mortality were recurrent or relapsed disease status (RR 3.05, p = 0.034), bacteremia (RR 3.1, p = 0.047), and need for ICU-level care as defined by intubation or pressor requirements (RR 8.2, p <0.001). 8 of 14 episodes requiring ICU-level care resulted in death. Severe malnutrition had a near-significant association with mortality (RR 2.7, p = 0.06). Gram-positive versus gram-negative status, bacterial resistance pattern, degree of neutropenia, type of malignancy, line of therapy, age, and duration of neutropenia were not significantly associated with mortality. There was no association between the presence of bacteremia and length of stay.Conclusion: Infection sources, in particular bacteremia, are identified with high frequency in leukemia patients experiencing NF. This study yielded generalizable outcome estimates for NF patients Risks of morbidity and mortality are high, with a 14% risk of ICU-level therapy and a 14% risk of mortality. Mortality risk increases in the setting of bacteremia, relapsed/refractory disease, and possibly poor nutritional status. The results suggest that early identification of correctable factors and rapid intervention during infections are essential to the management of high-risk hematologic malignancy patients. Early use of myeloid growth factor support may be worthy of study in selected high-risk populations.