Study design and patient population
This retrospective matching cohort study (1:2) was conducted at eight intensive care units (ICUs) between January 1st 2018, and January 1st 2020. A total of 132 eligible adult patients were included. This study was approved by the Medicine Institutional Review Board of Shanghai Jiaotong University, School of Medicine, Ruijin Hospital. The informed consent was waived.
The inclusion criteria were the following: age 18–80 years old; infected patients were confirmed with monomicrobial VAP caused by XDR E. coli or K. pneumoniae or P. aeruginosa or A. baumannii; at least two consecutive samples on different days (time interval at least 24 h) showed the presence of XDR GN bacilli from bronchial secretions or bronchoalveolar lavage samples; at least 6 doses of inhaled or intravenous polymyxin B usage. Patients who were younger than 18 years or older than 80, had multi-microbial VAP, cystic fibrosis or lung transplantation, were excluded.
Groups and matching criteria
Patients were divided into two groups: IV polymyxin B group and IH + IV polymyxin B group. All eligible patients were given at least 3 days of intravenous polymyxin B therapy. In addition, the IH + IV polymyxin B group received at least 6 doses of inhaled polymyxin B. To avoid the severe side effects caused by inhaled polymyxin B, inhaled glucocorticoid and bronchodilator were given 30 min before treatment. Furthermore, a vibrating mesh nebulizer was chosen to improve the atomization performance of inhaled polymyxin B. The practice of aerosol delivery of polymyxin B in the department of critical care medicine, Ruijin hospital north was according to Ehrmann et al. protocol . Standard atomization process was applied in six centers. Moderate-to-deep sedation was administered before starting nebulization. Vibrate mesh nebulizer was placed in the upstream of inspiratory limb, volume-controlled mode was chosen with constant inspiratory flow. End-inspiratory pause was set of 20% of duty cycle. Humidification system was removed during inhalation and resumed at the end of nebulization. Expiratory filter was changed after each nebulization. Patients in the two treatment groups were matched based on the day of data of initiation of polymyxin B therapy.
The primary endpoint of the present study was clinical cure of VAP. The favorable clinical outcome, microbiological outcome, VAP-related mortality, all-cause in-hospital mortality, VAP-related mortality within 28 days after initiation of polymyxin B, and side effects related to polymyxin B were selected as secondary endpoints. The primary outcome and all-cause in-hospital mortality were also assessed in subgroups (medical vs. surgical; low vs. high sequential organ failure assessment (SOFA)score; different responsible pathogens; with vs. without bacteria, with vs. without immunosuppressive status).
The following data were extracted from medical records: age, sex, history of antibiotic and glucocorticoids use within 30 days, complications, etiology, SOFA score, Acute Physiology, and Chronic Health Evaluation (APACHE II) score, microbiological diagnosis of VAP, antimicrobial susceptibility, dose and duration of polymyxin B therapy, duration of ICU stay, duration of mechanical ventilation, clinical outcomes, microbiological outcomes, and drug-related side effects. Data for VAP-related mortality and all-cause mortality during hospitalization, and VAP-related 28-day mortality were also extracted.
VAP was confirmed if pneumonia occurred more than 48 h after endotracheal intubation and mechanical ventilation . Pneumonia was defined based on clinical suspicion (with at least two of the following clinical infectious evidence): new lung infiltration confirmed by chest radiological examination (X-ray or CT scan) the new onset of fever; an oral temperature > 38 °C or < 36 °C; peripheral white blood cell count > 12 × 109/L or < 4 × 109/L, with or without nuclear-left shift; new-onset or aggravation of existing respiratory symptoms, such as purulent sputum, with or without chest pain and confirmed by a positive quantitative results for a respiratory sample (significant threshold ≥ 104 colony-forming units [CFU]/ ml) for bronchoalveolar lavages, ≥ 106 CFU/ml for endotracheal aspirations and ≥ 107 CFU/ml for noninvasive sputum samples). Septic shock was defined based on the Sepsis-3 definition .
The microbiological diagnosis was identified using regular biochemical methods. In vitro antimicrobial susceptibility tests were performed based on the criteria adopted at the Clinical and Laboratory Standards Institute. The minimal inhibitory concentration (MIC) was used to evaluate drug resistance. XDR Gram-negative bacilli were defined as bacilli that were only susceptible to polymyxins, including E.coli, K. pneumoniae, P. aeruginosa, and A. baumannii based on laboratory testing .
Clinical outcome was classified as clinical cure (i.e., the disappearance of infection-related symptoms and signs by the end of polymyxin B treatment), clinical improvement (i.e., improved infection-related symptoms and signs by the end of polymyxin B treatment compared with before polymyxin B treatment), clinical failure (i.e., persistence or worsening of presenting infection-related symptoms and/or signs during polymyxin B treatment or died). Recurrence of infection was defined as the emergence of infection symptoms (e.g., fever, shortness of breath), laboratory measures indicative of bacterial infection (e.g., C-reactive protein, procalcitonin, white blood cell count) within 72 h after discontinuation of polymyxin B, in the absence of other infection foci. Favorable clinical outcome included clinical cure or clinical improvement, and unfavorable clinical outcome included clinical failure or recurrence . Two physicians, who were blinded to the treatment, independently analyzed clinical outcomes. The reviewers evaluated the data again to achieve a consensus decision if disagreements about clinical outcomes occurred (about 8% of patients).
Bacteriological outcome was defined as elimination of the target XDR GN bacilli (i.e., no growth of the target pathogen in the lower respiratory specimens before discharge), persistence of the target XDR GN bacilli (i.e., persistent existence of the target pathogen with infection-related symptoms and signs), colonization (i.e., persistent existence of the target pathogen without infection-related symptoms and signs), and recurrence of the target XDR GN bacilli (i.e., regrowth of the same pathogen with infection-related symptoms and signs after discontinuation of polymyxin B during hospitalization) .
VAP-related mortality was defined as death due to the deterioration of signs of pneumonia and as death due to respiratory failure or septic shock.
Nephrotoxicity is the main side effect caused by the systematic administration of polymyxin B. Diagnosis of nephrotoxicity was based on risk, injury, failure, loss, end-stage kidney disease (RIFLE) criteria . At least over 1.5-fold increase in serum creatinine levels or a 25% decrease in calculated creatinine clearance from baseline and urine output less than 0.5 ml/kg/h over 6 h caused by intravenous polymyxin B was defined as nephrotoxicity. Proteinuria, tubular urine or azotemia caused by intravenous polymyxin B was also considered as nephrotoxicity. Skin hyperpigmentation has been also reported in the patients with polymyxin B treatment . The main adverse event of inhaled polymyxin B is bronchospasm.
Data were analyzed using SPSS software, version 22.0 (SPSS). Matched analysis was used. The matching criteria were as follows: age (± 5 years), gender, septic shock, and Apache II score (± 4 points) when started using polymyxin B.
Mean ± standard deviation was used to describe a continuous variable with a normal distribution and Student's t test was used for analysis. Median (interquartile range [IQR]) was used to describe continuous variables without normal distribution and rank-sum test was used to analyze. Categorical variables are expressed as frequency or ratio and analyzed using the χ2 test. When appropriate, results were reported as odds ratios (ORs) with associated 95% confidence intervals (CIs). P < 0.05 was considered statistically significant. The differences in VAP-related mortality between the IV group and IH plus IV group were assessed using Kaplan–Meier curves. Statistical significance was calculated using the log-rank test of the difference survival curves.
Multivariate logistic regression was used for analyze the effects of polymyxin B usage on the favorable clinical outcomes. The model was generated on the variables with clinical interest and on those with a p value of less than 0.05 in the univariate analysis. The included variables were: age, comorbidities, SOFA score, septic shock, daily dose of IV polymyxin B, treatment with IH + IV polymyxin B and duration of polymyxin B therapy. The effects of polymyxin B usage on the favorable clinical outcomes and on the all-cause of in-hospital mortality were also explored in the following subgroups: SOFA ≥ 8 vs SOFA < 8; surgery admission vs. medical admission; with vs without bacteremia; XDR A. baumannii vs. XDR E. coli vs XDR K. pneumoniae vs XDR P. aeruginosa; with vs. without immunosuppressive status. The cut-off value for SOFA score was determined according to the median value of our patients.