ABSTRACT
Carbapenems are currently the preferred agents for the treatment of serious Acinetobacter infections. However, whether cefepime-cefpirome can be used to treat an Acinetobacter bloodstream infection (BSI) if it is active against the causative pathogen(s) is not clear. This study aimed to compare the efficacy of cefepime-cefpirome and carbapenem monotherapy in patients with Acinetobacter BSI. The population included 360 patients with monomicrobial Acinetobacter BSI receiving appropriate antimicrobial therapy admitted to four medical centers in Taiwan in 2012 to 2017. The predictors of 30-day mortality were determined by Cox regression analysis. The overall 30-day mortality rate in the appropriate antibiotic treatment group was 25.0% (90/360 patients). The crude 30-day mortality rates for cefepime-cefpirome and carbapenem therapy were 11.5% (7/61 patients) and 26.3% (21/80 patients), respectively. The patients receiving cefepime-cefpirome or carbapenem therapy were infected by Acinetobacter nosocomialis (51.8%), Acinetobacter baumannii (18.4%), and Acinetobacter pittii (12.1%). After adjusting for age, Sequential Organ Failure Assessment (SOFA) score, invasive procedures, and underlying diseases, cefepime-cefpirome therapy was not independently associated with a higher or lower 30-day mortality rate compared to that with the carbapenem therapy. SOFA score (hazard ratio [HR], 1.324; 95% confidence interval [CI], 1.137 to 1.543; P < 0.001) and neutropenia (HR, 7.060; 95% CI, 1.607 to 31.019; P = 0.010) were independent risk factors for 30-day mortality of patients receiving cefepime-cefpirome or carbapenem monotherapy. The incidence densities of 30-day mortality for cefepime-cefpirome versus carbapenem therapy were 0.40% versus 1.04%, respectively. The therapeutic response of cefepime-cefpirome therapy was comparable to that with carbapenems among patients with Acinetobacter BSI receiving appropriate antimicrobial therapy.
INTRODUCTION
Bloodstream infections (BSIs) caused by Acinetobacter species are associated with considerable morbidity, mortality (29.7% to 36.9%) (1, 2), and health care costs (3) globally (4–6). Appropriate antimicrobial therapy is an important life-saving measure in patients with Acinetobacter BSI (7, 8). However, there are only a few studies on the efficacy of different appropriate antimicrobial regimens for the treatment of patients with Acinetobacter bacteremia. Carbapenems (imipenem, meropenem, or doripenem) and zwitterionic cephalosporins (cefepime or cefpirome) can be used to treat infections caused by Gram-negative bacteria susceptible to those antibiotics, including Acinetobacter species (9–11). However, the therapeutic efficacy of cefepime or cefpirome in severe Acinetobacter infection has not been fully evaluated. The antibiotic stewardship program (ASP) calls for reduced carbapenem use and increased diversity in administered antibiotics to reduce selective pressure caused by carbapenems leading to the emergence of carbapenem-resistant pathogens. Cefepime-cefpirome may be an alternative therapeutic regimen if these drugs are effective against Acinetobacter infections.
This retrospective study investigated the efficacy of cefepime-cefpirome versus carbapenem monotherapy administered within 48 h of bacteremia onset, as well as the impact on the 30-day mortality, of patients admitted to four medical centers in Taiwan with genomically identified monomicrobial Acinetobacter bacteremia after adjusting for underlying diseases and infection severity.
RESULTS
During the 6-year study period, 941 patients were found to meet the criteria for monomicrobial Acinetobacter bacteremia. They were derived from a population of 1,206 patients who had at least one Acinetobacter-positive blood culture. We excluded from the study 265 (22.0%) patients with polymicrobial bacteremia. Among the 941 patients with monomicrobial Acinetobacter bacteremia, 360 patients received appropriate antibiotic treatment within 48 h after bacteremia onset, with an overall 30-day mortality rate of 25.0% (90/360 patients). Among the 360 patients, 61 and 80 patients received cefepime-cefpirome and carbapenem monotherapy, respectively. The 141 patients receiving cefepime-cefpirome or carbapenem therapy were infected by Acinetobacter nosocomialis (51.8%), A. baumannii (18.4%), and A. pittii (12.1%). The demographic and clinical characteristics of the two groups are shown in Table 1. Patients in the cefepime-cefpirome group were more likely to have received chemotherapy. Those in the carbapenem group had significantly higher SOFA scores, had more central venous catheter and ventilator use, and were more likely to have had recent intensive care unit (ICU) admission and shock. There was no difference in the distributions of bacterial species between the two groups. The only significant difference in the source of bacteremia between the two groups was that from the respiratory tract in the carbapenem group. The durations of fever in those receiving carbapenem and cefepime-cefpirome therapy were 3.9 ± 1.6 (range, 0 to 9) days and 3.6 ± 1.5 (range, 0 to 7) days, respectively (P = 0.294). The antimicrobial susceptibilities and MIC distribution of Acinetobacter isolates in the cefepime-cefpirome and carbapenem treatment groups are shown in Table 2. The antimicrobial susceptibility rates of most antibiotics against Acinetobacter isolates in the carbapenem group were greater than 80%, except for those of ciprofloxacin (CIP), levofloxacin (LVX), and piperacillin-tazobactam (TZP). In the cefepime-cefpirome group, the antimicrobial susceptibility rates of all antibiotics against Acinetobacter isolates were greater than 80%, except for that of TZP.
Demographics and characteristics of patients with Acinetobacter bacteremia treated with cefepime-cefpirome versus carbapenem therapy
Antimicrobial susceptibilities and MIC distribution of Acinetobacter isolates in the cefepime-cefpirome and carbapenem treatment groups
The unadjusted 30-day mortality rates for cefepime-cefpirome and carbapenem therapy were 11.5% (7/61) and 26.3% (21/80), respectively. There are only 26 cases that were infected with A. baumannii and received cefepime-cefpirome or carbapenem therapy. The 30-day mortality rate was 50% (7/14) in the carbapenem group and 8.3% (1/12) in the cefepime-cefpirome group (P = 0.036). The factors identified in Cox regression analysis that significantly predicted 30-day mortality in patients receiving cefepime-cefpirome or carbapenem monotherapy are shown in Table 3. Multivariable analysis using propensity score as a covariate revealed that SOFA score and neutropenia were independent predictors of mortality (P < 0.001 and 0.010, respectively). After adjusting for age, SOFA score, invasive procedures, and underlying diseases, carbapenem versus cefepime-cefpirome therapy was not independently associated with higher or lower 30-day mortality rates (P = 0.110). The Cox regression analysis of the inverse probability of treatment weighting (IPTW) cohort showed similar results. SOFA score and neutropenia remained the independent prognostic factors for 30-day mortality but not carbapenem or cefepime-cefpirome therapy. To evaluate if SOFA score was an effect modifier, we split the cohort into two groups, one group with a SOFA score of ≤4 and a second group with a SOFA score of >4, and reran the analysis to see if mortality rates in patients who received cefepime-cefpirome versus carbapenem are different depending on SOFA group. In the group with a SOFA score of ≤4, the 30-day mortality rates of patients receiving carbapenem and cefepime-cefpirome therapy were 4.3% (1/23) and 6.9% (2/29), respectively (P = 1.000). In the group with a SOFA score of >4, the 30-day mortality rates of patients receiving carbapenem and cefepime-cefpirome therapy were 35.1% (20/57) and 15.6% (5/32), respectively (P = 0.050).
Cox regression analysis of the prognostic factors associated with 30-day mortality among patients with Acinetobacter bacteremia receiving carbapenems or cefepime-cefpirome therapya
Among the 360 patients who received appropriate antimicrobial therapy, SOFA score and neutropenia remained independent risk factors for 30-day mortality (see Table S1 in the supplemental material). Furthermore, the efficacy of cefepime-cefpirome therapy was as good as that for the other appropriate antimicrobial agents (hazard ratio [HR], 0.656; 95% confidence interval [CI], 0.284 to 1.518; P = 0.325) (Table S1).
We further analyzed the incidence density of mortality among patients with Acinetobacter bacteremia receiving cefepime-cefpirome or carbapenem monotherapy (Table 4). The incidence density of 30-day mortality in the cefepime-cefpirome group was lower than that of the carbapenem group (0.40% versus 1.04%, respectively) and was also lower than that of patients receiving appropriate antimicrobial agents other than cefepime-cefpirome and carbapenems (0.40% versus 1.17%, respectively).
Incidence density of mortality among patients with Acinetobacter bacteremia receiving appropriate antimicrobial therapy
DISCUSSION
To our knowledge, this is the largest multicenter study to evaluate the efficacy of cefepime-cefpirome and carbapenem monotherapy, to which the causative pathogen was susceptible in vitro, for the treatment of patients with monomicrobial Acinetobacter bacteremia after adjusting for multiple risk factors. We observed comparable therapeutic responses between cefepime-cefpirome and carbapenem therapies.
No comparative studies have investigated the therapeutic effects of different appropriate antibiotic treatments in Acinetobacter infections. Cefepime is frequently administered as empirical initial treatment in patients with febrile neutropenia after chemotherapy for hematological malignancy and solid tumor, which is why patients in the cefepime-cefpirome group were more likely to have received chemotherapy in this study. Cefepime as monotherapy or in combination with aminoglycoside antibiotics is used for coverage against multidrug-resistant Gram-negative bacteria (12, 13). In addition to febrile neutropenia, high-dose cefepime (2 g every 8 h administered over a 30-min infusion) had good efficacy in Gram-negative bacteremia-related meningitis and sepsis (14, 15). In our study, the incidence density of mortality was lower in the cefepime-cefpirome group than that in the carbapenem and other appropriate antibiotic therapy groups. These results indicated that cefepime-cefpirome can also be used to treat Acinetobacter BSI if the causative pathogen is susceptible to cefepime-cefpirome.
In our study, SOFA score and neutropenia were independent prognostic factors associated with 30-day mortality among patients with Acinetobacter bacteremia receiving appropriate therapy. Higher SOFA scores are related to mortality at the onset of Acinetobacter calcoaceticus-A. baumannii complex bacteremia (16–19). Furthermore, neutropenia is a predictor of mortality because innate immunity and the adaptive immune response play important roles in combating A. calcoaceticus-A. baumannii complex infections (20, 21). Neutropenia exacerbated A. calcoaceticus-A. baumannii complex infection in a murine model, and neutropenia and A. calcoaceticus-A. baumannii complex infection also themselves altered the release of proinflammatory cytokines, which can lead to serious bacterial disease (22).
The strengths of this study are the large number of patients from multiple medical centers in representative regions of Taiwan with various severities of infection using stringent inclusion criteria, recent isolates, and a well-defined endpoint of 30-day mortality. The use of cefepime-cefpirome or carbapenems in these four medical centers requires evaluation by an infectious physician to select empirical antibiotics. However, this was also the bias of this study. Cefepime was used empirically in febrile neutropenia, while carbapenems were used in patients with shock, pneumonia, and ventilator use. Further randomized controlled studies are warranted to evaluate whether appropriate cefepime treatment is equivalent to carbapenem or other appropriate antibiotic treatment.
The major limitations of this study are its retrospective design and the above-mentioned selection bias. Since the limited number of cases prevented subgroup analysis or matching, we analyzed the confounders that affected prognosis by Cox regression analysis. In addition, because the proportion of A. baumannii strains is relative low in this study, the conclusion is applied to patients with Acinetobacter bacteremia and not to those with A. baumannii bacteremia. Among patients with A. baumannii bacteremia, the 30-day mortality rate of the cefepime-cefpirome therapy group is lower than that of the carbapenem group, but the case number is small (n = 26). We will continue to collect data from patients treated with cefepime-cefpirome and carbapenem therapy for A. baumannii bacteremia to see if the finding is consistent.
In conclusion, among patients with Acinetobacter BSI, the efficacy of appropriate antibiotic treatment with cefepime-cefpirome was comparable to that with carbapenems. Further randomized controlled studies are required to assess the therapeutic effect of cefepime-cefpirome in cases of severe Acinetobacter infection.
MATERIALS AND METHODS
Study population.This retrospective study was conducted from January 2012 to August 2017 at four medical centers (Changhua Christian Hospital [CCH], MacKay Memorial Hospital [MMH], Tri-Service General Hospital of National Defense Medical Center [TSGH], and Taipei Veterans General Hospital [TVGH]) in Taiwan. The charts were reviewed for all patients with at least one positive blood culture for Acinetobacter species with symptoms and signs of infection, such as fever (body temperature, ≥38°C), chills, or hypotension. One patient included only one positive blood culture. For patients with two or more positive blood cultures, only the first blood culture was included. This study included patients who received appropriate antibiotics, defined as the administration of at least one antimicrobial agent to which the causative pathogen was susceptible in vitro within 48 h after bacteremia onset, with an approved route and dosage appropriate for end organ function for at least 48 h. The carbapenems used included imipenem, meropenem, or doripenem. Bacteremia onset was defined as the day on which the blood culture that eventually yielded Acinetobacter species was drawn. The primary outcome measurement was all-cause 30-day mortality. The study protocol was approved by the hospitals’ institutional review boards (IRBs) (CCH, IRB no. 140514; MMH, IRB no. 14MMHIS125; TSGH, IRB numbers 1-103-05-100 and 1-105-5-100; and TVGH, IRB numbers 2014-07-006CC and 2015-04-003C).
Microbiological assessment.The presumptive identification of the isolates to the level of Acinetobacter species was determined using a Vitek 2 system (bioMérieux, Marcy l’Étoile, France). A. baumannii was identified using multiplex PCR (23). Isolates identified as non-baumannii Acinetobacter spp. were further genomically identified to the species level by 16S-23S ribosomal DNA intergenic spacer sequence analysis, as previously described (24). The MICs for the antibiotics administered were determined using a Vitek 2 (bioMérieux) automated system. The results were subsequently interpreted according to Clinical and Laboratory Standards Institute (CLSI) guidelines (25).
Data collection and definitions.Patient medical records were reviewed to obtain clinical information, such as demographic characteristics, underlying diseases, duration of hospital stay, time of receipt, dose, and route of administration of individual antimicrobials, and use of invasive procedures at the time of Acinetobacter bacteremia onset. Shock was defined as persisting hypotension requiring vasopressor therapy to elevate mean arterial pressure of ≥65 mm Hg and having a lactate concentration of >2 mmol/liter (18 mg/dl) despite adequate fluid resuscitation occurring within two days after drug treatment was started. Chemotherapy was defined as the administration of cytotoxic agents within 6 weeks before bacteremia onset. Immunosuppressive therapy was defined as the use of immunosuppressive agents within 2 weeks or the use of corticosteroids at a dosage equivalent to or higher than 15 mg of prednisolone daily for 1 week within 4 weeks before bacteremia onset. We defined moderate-to-severe renal disease as an estimated glomerular filtration rate of <50 ml/min/1.73 m2. Hemodialysis was defined as receipt of hemodialysis within three days before the day of positive blood culture. Chronic lung diseases included chronic obstructive pulmonary disease, asthma, bronchiectasis, pulmonary fibrosis, and old pulmonary tuberculosis. The source of bacteremia was determined according to the definitions of the U.S. Centers for Disease Control and Prevention (26). Polymicrobial bacteremia was defined as the isolation of one or more microorganisms other than Acinetobacter species from blood during the same bacteremia episode. Infection severity was evaluated based on patient SOFA score (27) within 24 h prior to bacteremia onset.
Statistical analysis.The baseline demographic characteristics of patients with Acinetobacter bacteremia treated with either cefepime-cefpirome or carbapenems were compared by Pearson χ2 and Wilcoxon rank sum tests for categorical and parametric continuous variables, respectively. A Cox proportional hazard regression model was used to explore the independent prognostic factors associated with 30-day mortality. All biologically plausible variables with P values of <0.10 in the univariable analysis were considered for inclusion in the Cox regression model in the multivariable analysis. P values of <0.05 indicated statistical significance.
To reduce the effects of confounding, further analyses were performed using propensity scores. Age, sex, and covariates from the primary analysis, including recent ICU admission within 1 month, SOFA score, receipt of chemotherapy, neutropenia, type 2 diabetes mellitus, central venous catheter use, ventilator use, and respiratory tract infection as the source of bacteremia were used to calculate the propensity score using a logistic regression model with the carbapenem treatment group as the dependent variable. The model obtained had an area under the receiver operating characteristic curve of 0.785. The propensity score was used in two further analyses. At first, the propensity was used as a covariate in multivariate Cox regression analysis. While the propensity score matching eliminates a substantial number of cases from the analysis, an IPTW analysis was conducted. Weighted Cox regression analysis was performed to evaluate the prognostic factors associated with 30-day mortality.
All statistical analyses were conducted using PASW Statistics for Windows, version 26.0 (SPSS, Chicago, IL, USA).
ACKNOWLEDGMENTS
This work was supported by grants from the National Yang-Ming University Hospital (grant RD2019-011), Taipei Veterans General Hospital (grants V106B-002, V107C-037, V108C-012, VTA106-T-5-3, VTA107-T-3-2, VTA108-T-2-2, and VTA108-T-2-3), Tri-Service General Hospital of National Defense Medical Center (grants TSGH-C107-099, TSGH-C108-137, TSGH-C109-144, MAB-107-095, and MAB-108-038), and the Ministry of Science and Technology (grants MOST 105-2314-B-016-039-MY3, MOST 107-2314-B-075-066-MY3, MOST 107-2314-B-016-051-MY3, and MOST 108-2314-B-016-029).
The funders had no role in the study design, data collection and interpretation, or the decision to submit the work for publication.
We declare no conflicts of interest.
The members of the ACTION study group include Yea-Yuan Chang (National Yang-Ming University Hospital, Yilan, Taiwan), Ya-Sung Yang (Division of Infectious Diseases and Tropical Medicine, Department of Internal Medicine, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan), Chung-Ting Chen (Department of Emergency Medicine, Taipei Veterans General Hospital, Taipei, Taiwan), Yuag-Meng Liu (Changhua Christian Hospital, Changhua, Taiwan), Shu-Chen Kuo (National Institute of Infectious Diseases and Vaccinology, National Health Research Institute, Maoli County, Taiwan), Chang-Pan Liu (MacKay Memorial Hospital, Taipei, Taiwan), Te-Li Chen (Graduate Institute of Life Sciences, National Defense Medical Center, Taipei, Taiwan), and Yi-Tzu Lee (Taipei Veterans General Hospital, Taipei, Taiwan).
Y.-Y.C., Y.-S.Y., and Y.-T.L. conceived the project and wrote the manuscript. Y.-Y.C., Y.-S.Y., Y.-T.L., and T.-L.C. designed the experiments and made manuscript revisions. Y.-Y.C., Y.-S.Y., and the ACTION Study Group collected the clinical data. Y.-C.W. and Y.-T.L. performed the experiments. Y.-Y.C., Y.-T.L., and S.-L.W. conducted the statistical analysis. All authors reviewed the manuscript.
FOOTNOTES
- Received 3 December 2019.
- Returned for modification 5 February 2020.
- Accepted 8 March 2020.
- Accepted manuscript posted online 16 March 2020.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.
