Bloodstream bacterial infections are a potentially life-threatening complication of cytotoxic chemotherapy, with approximately 10 to 25% of febrile neutropenic (FN) patients having a positive blood culture taken at the time of first fever during neutropenia (Liu et al ., 2011; Gustinett, Mikulska 2016). Empiric antibiotic therapy started promptly at onset of fever and neutropenia will reduce morbidity and mortality and is standard of care. The choice of empiric antibiotics depends on the prevailing bacterial pathogens in a given at-risk population. Accordingly, it is essential know the patterns of bacterial types and antibiotic susceptibilities in the population at risk, especially in the global context of rising bacterial resistance to common antibiotics. Currently, no such data exists from the United States; a national survey of bloodstream isolates from cancer patients was last published in 2003(Wisplinghoff et al ., 2003). The BISHOP study aims to generate a geographically broad data set describing bloodstream isolate epidemiology, antibiotic susceptibility and clinical outcomes of bacteremias in patients with hematologic malignancies (HM) or undergoing allogeneic or autologous HSCT in the United States.


Fifteen geographically diverse US cancer centers that perform at least 100 HSCTs yearly are participating, starting from December 2016. Patients are consecutively enrolled at each center if they have a documented bacterial bloodstream infection at the time of first fever and neutropenia within 14 days of receiving cytotoxic chemotherapy for treatment of HM or in preparation for HSCT. Bacterial isolates are shipped to the central lab (UNMC) and frozen; microtiter antibiotic susceptibility testing will be performed in the future. Clinical data collected electronically include patient demographics, clinical course over first 24 hours, empiric antibiotic treatments and outcomes at 30 days. Target collection is 720 bacterial isolates over 18 months.


Eleven of 15 participating centers enrolled 187 patients (chemo only 127, HSCT 60), with 213 bacterial bloodstream isolates in the first 8 months of study (Table 1). There were 49% gram negative isolates, with E.coli (22%), Klebsiella pneumoniae (7.5%) and Pseudomonas aeruginosa (7.5%) predominating. Among gram positives (46%) , Viridans group streptococci (21.6%), coagulase negative staphylococci (7.0%) and methicillin-resistant Staph aureus (MRSA) (5.2%), predominated. Anaerobes accounted for 5% of isolates. Table 2 shows types of bacteremias in the 187 patients, with about equal proportions of single gram negative and gram positive isolates. Full susceptibility data have not yet been collated. Ninety seven patients (chemo only 63, HSCT: 19 allo,15 auto) have clinical data available for the initial 24 hours after presentation: 39% were female, median age 59 years (23-73 y), median ANC 0 cells/ul (0-430). Underlying primary diagnoses were AML 59%, NHL 12%, MM 10%, ALL 6%, MDS 3% HL 2%, and other 7%. Antibiotic prophylaxis prior to onset of FN was given in 61%. Complicated presentations were mucositis 29%, pneumonia 13%, altered mental status 7% or hypotension/pressor requirement 9%. Empiric antibiotics included cefepime 58%, pip-tazobactam 19% ,meropenem 10%, ceftazidime 4%, aztreonam 4%, doripenem (1%) and Other 4 %. Vancomycin and tobramycin were included in the initial regimen in 41% and 3% respectively. Median MASCC score was 21 (6-26). Median Pitt bacteremia score was 1.0 (0-7). There were no deaths within the first 24 hours after presentation.


The initial data from the BISHOP survey is described here, representing the largest multicenter profile of bloodstream infections among patients with HM or HSCT collected in the US. Gram negative and positive organisms are distributed equally, with E.coli and viridans group streptococci most commonly isolated. MASCC and Pitt score on presentation did not indicate that this patient cohort was at high risk for early complications (first 24h). Susceptibility data, when available, may inform empiric antibiotic choices. Notably, no patients died within the first 24 hours after presentation indicating that current empiric therapy strategies are generally successful in the early period of fever and neutropenia.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.