Clinical treatment of hospital-acquired pneumonia caused by extensively drug-resistant Acinetobacter baumannii: single-centre retrospective study

Changquan Fang Huizhou Central People's Hospital Limin Xu Huizhou First People’s Hospital Junhong Lin Huizhou Central People's Hospital Yujun Li Guangzhou First Municipal People’s Hospital, South China University of Technology Shuquan Wei Guangzhou First Municipal People’s Hospital, South China University of Technology Zhuxiang Zhao Guangzhou First Municipal People’s Hospital, South China University of Technology Ziwen Zhao (  zhaozw@yeah.net ) Guangzhou First Municipal People’s Hospital, South China University of Technology

especially extensively drug-resistant A. baumannii (XDRAB), in hospital infections has been increasing annually [2,3], posing a signi cant clinical challenge in anti-infective treatment.
XDRAB refers to strains that are resistant to current commonly used antibacterial drugs in clinical settings and are only sensitive to one to two drugs with potential anti-Acinetobacter activity [4]. Since the rst XDRAB isolate was reported in Taiwan in 1998, these strains have broadly spread worldwide. XDRAB infections are particularly common in critically ill patients and in those with severe underlying diseases, resulting in a high mortality rate [5].
At present, the optimal treatment strategy for XDRAB infection remains controversial, and the drugs that are available for treating XDRAB are also extremely limited [6]. Once XDRAB infection occurs, especially lower respiratory tract infection, the prognosis of patients is often poor. Therefore, it is important to identify the clinical factors that contribute to infection and prognosis among patients with hospitalacquired XDRAB infections.
Toward this end, we performed a retrospective analysis of the clinical features and treatment outcomes of patients clinically diagnosed with XDRAB pneumonia at our hospital from January 2016 to December 2017.

Study subjects
This study was designed as a retrospective case analysis. Cases of hospital-acquired pneumonia caused by XDRAB clinically diagnosed at Guangzhou First Municipal People's Hospital from January 2016 to December 2017 were reviewed. Inclusion criteria were as follows: (1) XDRAB was isolated and cultured from sputum, breoptic bronchoscopy drainage uid, lavage uid, or venous blood; (2) typical manifestations of bacterial infection (e.g. fever, and increase in white blood cell count, C-reactive protein, and procalcitonin); (3) clinical symptoms and signs consistent with pneumonia; and (4) imaging manifestations of new, continuous, or worsened lung exudation, in ltration, or consolidation [7]. The exclusion criteria were: (1) aged < 18 years; (2) XDRAB was cultured not less than twice; (3)

Statistical analysis
Attributes data are expressed as frequency (number, percentage) and were compared between groups using the χ 2 test. Normally distributed data are expressed as mean ± standard deviation, which were compared using Student's t-test, whereas non-normally distributed data were compared with Mann-Whitney U test.The factors affecting prognosis (survival or death) were identi ed in logistic multiple regression analysis. P < 0.05 was considered to indicate a statistically signi cant difference or association with prognosis.All statistical analyses were performed with Statistical Package for the Social Sciences,Version 19.0(SPSS 19.0).

Patient characteristics
A. baumannii was isolated from a total of 547 patients in the 2 years, 143 of which were identi ed as XDRAB for a detection rate of 26.1%. After excluding records of duplicates (10 cases), non-respiratory infections (25 cases), colonisation [7] (40 cases), XDRAB cultured not less than two times (11 cases), and incomplete clinical data (5 cases), a total of 52 cases of clinically diagnosed XDRAB pneumonia were retained. In addition, cases of mixed infections (3 cases) and non-pneumonia-related death (1 case) were excluded, leaving a total of 48 cases that met the inclusion criteria. The ow chart for patient selection is shown in Figure 1.
The mean age of the included patients was 78 ± 15 years, including 41 men and 7 women. All 48 patients with XDRAB pneumonia were admitted to the ICU, all received invasive mechanical ventilation and deep venous catheterisation, eight patients had a tracheotomy, and six patients had haemodialysis. Most of the patients had multiple underlying diseases, including 32 cases of hypertension, 15 cases of coronary heart disease, 26 cases of chronic cardiac insu ciency, 23 cases of chronic obstructive pulmonary disease, 11 cases of chronic renal insu ciency, 9 cases of malignant tumour, and 9 cases of diabetes. The most common source of the specimen was bronchoscopy drainage uid (26 cases), followed by sputum (14 cases) and venous blood (8 cases). Of the 48 patients, 20 (42%) survived and 28 died (58%). Table 1 shows a comparison of the antibacterial drugs used during XDRAB infection in relation to the prognosis of patients (survival versus death). There was no signi cant difference between groups with respect to the commonly used antibacterial drugs with potential antibacterial activity against A. baumannii or their combinations. All 48 patients exhibited chest imaging progression during the infection period, which most commonly manifested as multiple pulmonary lobar in ltrates and patchy exudation.

Associations of antibacterial use during infection with clinical characteristics
The time from the rst isolation of the pathogen to the improvement of chest X-ray absorption in the survival group was shorter (8.3 ± 4.7 days) than that of the death group (12.4 ± 12.0 days) the difference statistically signi cant Thirty-ve patients had fever during the infection, 16 of whom were in the survival group with a mean time from rst isolation of the pathogen to fever abatement of 10.2 ± 9.8 days, whereas the mean time from rst isolation to fever abatement or death for the 19 patients in the death group was 7.2 ± 7.4 days. There were no signi cant differences in the time of persistent fever between the two groups, although the frequency of patients with persistent fever for ≤3 days and for 4-7 days after infection was higher for the death group, whereas relatively more patients in the survival group had a persistent fever for >7 days ( Table 2).

Factors affecting bacteria clearance
Bacteria were effectively cleared in 20 of the 48 cases. As shown in Table 2, advanced age, number of organ failures, severity of illness (APACHE II score), and airway care (tracheotomy, sputum suction) signi cantly affected bacterial clearance.

Prognostic factors
XDRAB was cleared in all cases in the survival group and in no cases in the death group. As shown in Table 3, age and APACHE score emerged as independent risk factors affecting prognosis in multiple logistic regression; tracheotomy during infection was a signi cant protective factor. There was no effect of various combinations of drugs with anti-Acinetobacter activity on survival.

Discussion
A. baumannii has become an important pathogen of nosocomial infections, which can easily cause epidemics, especially in the ICU, and XDRAB has received particular attention in this regard [14]. A. baumannii was previously considered to be a low-virulence opportunistic pathogen with no impact on the prognosis of hospitalized patients [15]; however, recent studies have shown that A. baumannii, especially multidrug-resistant strains, are commonly isolated in critically ill patients, and are associated with a high mortality rate ranging from 52% to 66% [16]. The mortality rate from XDRAB pneumonia in this study was 58.3%, which is similar to that reported by Boral et al. [17], con rming the high mortality rate of this pathogen. Therefore, it is particularly important to deepen the exploration of strategies for the treatment of XDRAB infection.
The rst challenge is the selection of an appropriate antibacterial drug for the treatment of XDRAB infection, which requires an extensive evaluation of the role of various drugs with potential anti-Acinetobacter activity on the treatment outcome. Commonly used antibacterial drugs that are currently recommended for the treatment of A. baumannii infections include sulbactam and a compound preparation of sulbactam-containing β-lactam antibiotics, carbapenem antibiotics, polymyxin antibiotics, tigecycline, tetracycline, aminoglycoside, and quinolone [16,18]. In cases of XDRAB infection, a two-drug or even a three-drug combination is often used. Although some studies have shown potential bene ts of combined regimens, these results are mostly based on animal experiments, in vitro studies, and uncontrolled clinical studies with a small number of cases, and the conclusions are inconsistent [19]. The possibly effective two-drug combination regimens are (1) sulbactam or a compound preparation containing sulbactam as the basis in combination with one of carbapenems, minocycline (or doxycycline), polymyxin, or aminoglycoside antibiotics [20][21][22]; and (2) tigecycline as the basis combined with a compound preparation containing sulbactam, aminoglycoside antibiotics, or quinolone [23-26]. The three-drug combination regimens include a compound preparation containing sulbactam (or sulbactam) combined with doxycycline and carbapenem antibiotics [20].
All of the recommended drugs with anti-Acinetobacter activity had been used on the patients included in this study with con rmed XDRAB infection, except for polymyxin antibiotics. This is because polymyxin has a limited effect due to its low blood concentration in the lung and cerebrospinal uid. Kim et al. [27] reported that the clinical response rate of patients with XDRAB pneumonia who received polymyxin or tigecycline as the basis treatment was 48% and 47%, respectively, with no signi cant difference. Yilmaz et al. [28] reported that the clinical and microbial treatment response rates for the single use of polymyxin, polymyxin combined with sulbactam, and polymyxin combined with carbapenem in the treatment of multidrug-resistant or XDRAB pneumonia were 63.6%, 55%, and 60%, respectively, with no signi cant differences. In a meta-analysis, Jung et al. [19] found no signi cant difference in the clinical and microbial treatment response rates of polymyxin or tigecycline alone, or in combination with carbapenem and sulbactam. Liu et al. [29] also reported no signi cant difference in the prognosis of patients with multidrug-resistant A. baumannii pneumonia who were treated with or without tigecycline. Consistently, in the present study, there was no signi cant difference in the use of various antibacterial drug combinations (including monotherapy, two-drug combinations, and three-drug combinations) during infection between the survival group and the death group.
Although the recommended anti-A. baumannii antibacterial drug combinations were used at our hospital, following these guidelines did not signi cantly improve the prognosis or bacterial clearance rate. There are several potential reasons to explain these observations. First, A. baumannii is a low-virulence pathogen. Therefore, even if extensive drug resistance appears, the toxicity would not necessarily be enhanced. Second, antimicrobial drugs cannot effectively clear XRDAB, and therefore other adjuvant treatments are required. Third, the prognosis may be mainly affected by factors other than treatment, such as the underlying condition.
Jung et al. [19] found that advanced age, multiple organ failure, and a severe disease condition had a negative effect on bacterial clearance; effective airway management (e.g. tracheostomy, sputum suction) was conducive to bacterial clearance; and the use of antibacterial drugs had no obvious effect on prognosis, which are all consistent with the present ndings. Multivariate logistic analysis showed that age and APACHE II score were independent risk factors affecting prognosis, and that tracheotomy is a protective factor.
Patients with advanced age, multiple organ failure, and severe disease conditions are mostly bedridden, and may even be in a coma and con ned to bed for a long time. This situation is typically accompanied by a decreased cough re ex, resulting in poor sputum-discharging and clearance ability of the respiratory tract [30]. Therefore, it is necessary to strengthen airway management, especially the body position and mechanical-assisted sputum discharge. For patients with thick sputum and weakness in expectoration, tracheotomy should be performed as soon as possible. It is currently believed that long-term tracheal intubation is more likely to cause airway damage, infection, and patient discomfort, and requires more doses of sedatives. Therefore, for patients who require long-term mechanical ventilation, tracheotomy should be performed as soon as possible to replace tracheal intubation [31][32], as tracheotomy results in a more stable arti cial airway. This also allows patients to eat by mouth, and tracheotomy is further conducive to the removal of pulmonary secretions, which can improve the overall prognosis [33][34][35]. Kimura [36] reported that mechanical ventilation with tracheotomy could effectively prolong the median survival of patients with lateral sclerosis. However, the optimal timing of tracheotomy remains controversial [35]. The National Association for Medical Direction of Respiratory Care recommends that patients who have been under tracheal intubation for more than 3 weeks should receive tracheotomy as a substitute [37]. At present, early tracheotomy is preferred. Timely tracheotomy can reduce the complications of long-term tracheal intubation (e.g. larynx injury, airway injury, bacterial growth) and reduce the rate of pulmonary infection, making the infection easier to be controlled, which is closely related to shorter hospital stays, lower treatment costs, and lower mortality [38][39][40].
Regarding the prognosis of drug-resistant A. baumannii infections, it has been reported that the virulence of the drug-resistant bacterium itself does not increase, and therefore neither would the mortality rate, and that it is instead the severity of the underlying disease that will ultimately affect the prognosis [1], in line with our results. At present, the APACHE II score is the most widely used and authoritative critical illness condition evaluation system in clinical ICU wards, which can provide an objective and scienti c basis for the rational use of medical resources and prognostic judgements [7]. However, the APACHE II scale is mainly designed for ICU inpatients and is not speci c to pneumonia, and few studies have directly assessed its prognostic value in patients with pneumonia. This study shows that an APACHE score ≥20 can indicate a poor prognosis in patients with XDRAB hospital-acquired pneumonia, similar to the ndings of Liu et al. [29].
This study also has some limitations, which should be mentioned. This was a retrospective study, we were lacking data on polymyxin treatment, and this was a single-centre study with a relatively small sample size. Therefore, further in-depth research using prospective multi-centre studies with a large sample size are needed.

Conclusions
XDRAB hospital-acquired pneumonia has a high mortality rate. Advanced age and a severe disease condition are independent risk factors for a poor prognosis. Although antibacterial drugs do not have a clear effect on improving prognosis during infection, XDRAB pneumonia is curable. In particular, effective airway management (tracheotomy, sputum suction) is the key to clearing XDRAB, thereby improving the prognosis of patients. The data used in our study was anonymized before its use. All the data were obtained from the medical records. All data generated or analyzed during this study are available from the corresponding author upon reasonable request.All patients and legally authorized representative/next of kin of deceased patients provided informed consent for this study.

Consent for publication
Not applicable.

Availability of data and materials
Our present study was a retrospective observational study. All the data were obtained from medical records of patients. The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request, but the identifying/con dential patient data would not be shared.  Figure 1 Flow chart of selection of patients with extremely drug-resistant Acinetabacter baumanii (XDRAB) infection for this study.