Molecular epidemiology of multidrug-resistant clinical isolates of Acinetobacter baumannii

Acinetobacter baumannii is an emerging opportunistic microorganism [1, 2] which may cause severe nosocomial infections [3‐13]. International spread of multidrug-resistant A. baumannii (MDR A. baumannii) has been described worldwide and international transfer of colonized patients has led to the introduction and subsequent epidemic spread of MDR A. baumannii strains from southern to northern European countries [14, 15].

Exposure to healthcare systems of countries with high-prevalence (HPCs) of multidrug-resistant microorganisms (MDRO) has been described as a risk factor for the spread and acquisition of these microorganisms [16‐18]. Although Austria is situated in the heart of the European Union with historic and current strong ties to East and Southeastern European countries, so far no epidemiologic investigation has been conducted to analyses possible international transmission of MDR A. baumannii in this geographic region. Austria hosts some 1600000 individuals (18% of the population) from East European member states, who regularly visit their home countries [19]. More than one third of these guests live in Vienna. The Vienna General Hospital (VGH), a large university teaching hospital, serves as specialist tertiary care center for severely ill patients from all parts of Austria as well as for patients from surrounding countries frequently admitted for burns or transplantation surgery.

The national antimicrobial resistance rate in Austria is low compared to the European average where in 2015 the overall percentage of invasive A. baumannii strains resistant to carbapenem was 6.4% [20]. From 2014 to 2016 the mean number of A. baumannii isolated at VGH was 130 per year. The yearly rates of MDRO were 25%. Beginning in 2007, the VGH registered an alarming increase of A. baumannii strains resistant to at least four antibiotic classes (penicillins, cephalosporins, fluoroquinolones and carbapenems), particularly in patients admitted to ICUs.

A total of 76 patients with microbiological positive test for MDR A. baumannii were recorded and of these 44 (56%) were isolated from skin and wounds, 24 (30%) from respiratory tract specimens, 5 (6%) from rectal swabs, 4 (5%) from central venous catheters and 2 (3%) from abscess aspirates of the patients. A total of 47 (62%) colonized and 29 (38%) infected patients were admitted to altogether 41 wards (13 medical wards, 8 surgical wards, 16 ICUs and 4 IMCs). Of the 76 patients, 50 (66%) were male and 26 (34%) female, 7 were children under 8 years old (6 male, 1 female) and the age of the overall patients ranged between 0 days and 78 years (mean 43.3 years, SD ± 21 years). The duration of the hospital stay at the VGH lasted between 7 and 186 days (mean 58 days, SD ± 42 days). All patients spent at least part of their hospital stay in ICUs (range: 4–186 days, mean stay 32 days, SD ± 33 days) and additionally 7 of them in an IMC (range: 2–162 days, mean stay 13 days, SD ± 16 days). Multiple trauma was the cause for the hospital stay in 29% (22/79) of patients, followed by acute lung diseases (20%; 15/76), burns (12%; 9/76), and cancer (9%; 7/76). Of the patients 15 died during the hospital stay and 8 (53%) thereof due to pneumonia and sepsis. By reviewing the clinical charts of these patients, a causal relationship between MDR A. baumannii infection and death was found in 6 cases.

The results of susceptibility testing showed that out of 79 A. baumannii isolates resistant to meropenem, 21 (27%) were additionally resistant to imipenem and 45 of the 79 isolates (62%) were moreover resistant to aminoglycosides. No isolate showed resistance to colistin. A significant trend of increasing incidence of MDR over the 9‑year period was found (Cochran-Armitage p < 0.05).

All 79 isolates were genotyped by the Diversilab® system. By using a similarity threshold of 95%, the analysis of the data identified 23 different genotypes, of which 13 were only observed once and showed a Dice similarity coefficient to all other strains between 81% and 92%. The 10 remaining genotypes were found repeatedly and they could be grouped into 10 clusters (I–X), consisting of a minimum of 2 to a maximum of 21 strains (Table 1). In cluster VI two strains belonged to the same person and differed in antibiotic susceptibility. Cluster VII consisted of two strains belonging to the same person but differed in antibiotic resistance. One patient had at the same time one strain belonging to cluster X and one strain observed only once, those showed different antibiotic susceptibility. All other strains of the 10 clusters belonged to different patients.

While with clusters II, IV, VII and X there was no overlap in the times of stay of all patients, with cluster III there was a close occurrence in the times of stay (3 months). With clusters V (28 months), VI (39 months), VIII (16 months) and IX (18 months) the time gaps up between the times of stay of the patients was far more high.

Of the MDR A. baumannii strains 56% (44/79) were first isolated in ICUs, 28% (22/79) in surgical departments, 15% (12/79) in medical and 1 in an IMC unit. Of the MDR A. baumannii strains 56% (44/79) were acquired before patient admission to the VGH and 44% (35/79) during their hospitalization. Regarding the nosocomial acquired A. baumannii strains, 6 index patients could be identified as source of secondary transmission. Within cluster I, this occurred between 2 patients after a 3 weeks time frame; within cluster II the time frame of acquisition between 2 patients was 10 days. Both transmissions occurred in ICUs. Of the 6 acquired A. baumannii documented within cluster III, the mean time frame was 93.2 days (±SD 89 days) and 71.4% (5/6) of them occurred in ICUs. Within cluster VI, A. baumannii time frame of acquisition between index patient and secondary case was 240 days, 11 days within cluster VIII and 8 days within cluster IX. All acquisitions within these three clusters occurred in ICUs. Of the 79 patients 31 (41%) had no contact with healthcare systems of HPCs for MDR A. baumannii in the 4 weeks before VGH admission, 77% (24/31) of them were Austrian citizens with no foreign background or travel history. Of the 76 patients 27 (36%) had a previous contact with a healthcare setting of an EU-HPC for MDR A. baumannii and 67% (18/27) of them with Rumanian hospitals. Within the patients that had previous contact with a healthcare setting of HPCs outside the EU (18/76; 24%), 13 (72%) were transferred from hospitals of the former Yugoslavia (Table 2). The country of origin of the patients and country distribution according to MDR A. baumannii prevalence are illustrated in Fig. 1.

Within the patients that acquired A. baumannii strains before admission to the VGH, previous contact with healthcare settings of HPCs for MDR A. baumannii was found in 71% (31/44). Within the patients that acquired A. baumannii after VGH admission this was 49% (17/35, χ²-test 3.9; p < 0.05) (Fig. 2).

Of 6 identified index cases belonging to the genotyped clusters, where A. baumannii was nosocomially acquired (I, II, III, IV, VI, VIII, IX), 5 (83.3%) had a previous contact to a healthcare system of a HPC for MDR A. baumannii: 2 from an EU country (Romania) and 3 from a country outside the EU (Serbia and Macedonia).

Our results show that ICUs are the most frequent setting for MDR A. baumannii in-hospital acquisition. Both colonized and infected patients were severely ill and spent all or part of their hospital stay in an ICU. Almost half of these cases were polytrauma patients or had suffered severe burns. The ICUs were also the setting were more than half of the patients belonging to the identified genotyped clusters that were cared for. These findings support the role of indwelling devices and invasive procedures for MDR A. baumannii colonization or infection [27].

Interestingly we found a long time frame between strain acquisitions in three genotyped clusters (VI, VIII and IX). We decided, however, not to split these clusters as the excessive time of MDR A. baumannii acquisition within these clusters may be due to the persistence of A. baumannii in a hospital environment and confirms the importance of contamination of hospital inanimate surfaces in the transmission of these pathogens [28]. This could be, under certain conditions, far longer than the 5 months reported in literature. Nevertheless, cases of A. baumannii colistin resistance have been increasingly reported in Europe [29]; however, no such isolate was found in our hospital in the last 9 years, which reflects the drug resistant microorganism rates in our country [20].

Because of the worldwide alarming trend in carpabenem and multidrug resistance, specific efforts of our epidemiological research was focused on the previous contact of the MDR A. baumannii patients with healthcare systems in other countries. Approximately two thirds (48/76) of the patients were exposed to a healthcare setting of HPCs for MDR A. baumannii in the previous 4 weeks before VGH hospitalization and the same proportion (31/44) was observed within those who acquired A. baumannii strains before admission in our hospital. Furthermore 5 out of 6 identified index patients were admitted to healthcare services of HPCs for MDR A. baumannii, confirming exposure to these healthcare settings as a major risk for A. baumannii colonization or infection.

We chose not to differentiate between type of healthcare exposure and length of hospitalization in such countries because no evidence was found in the literature to be a relevant determinant of MDRO acquisition [17]. Even if past studies used longer exposure periods to a healthcare system abroad [16, 29], we chose a 4-week time frame before admission to our hospital, in order to avoid possible errors in the collection of the patient hstory records.

Our study had some limitations. Firstly, we did not examine other recognized factors associated with MDR A. baumannii colonization and infection, such as the antibiotic use before hospitalization in our hospital. Then, as a systematic microbiological screening of fellow patients was not conducted during the observation period, the role of asymptomatically colonized patients in the transmission chain remained unknown. In his research Joly-Guillou [5] indicated a ratio of colonized patients to infected patients of 10:1 for Acinetobacter in a clinical unit and concluded that the patients with symptoms of Acinetobacter infection are likely to represent the tip of the colonization iceberg and significant undetected dissemination is also likely to be occurring. This suggests that in our case the colonized patients could have been far more than the 47 identified in our hospital. The strengths of our study were the long-time observation period and that not only invasive but also non-invasive isolates of A. baumannii were investigated. As far as we know, no other study has collected data on the molecular characterization of MDR A. baumannii strains of colonized and infected patients for such a long period of time.

In conclusion, the findings of our study warrant the implementation of a targeted risk-adjusted screening policy for A. baumannii for all patients before admission to ICUs. Given the high percentage of MDR A. baumannii cases that were imported from foreign countries in our study population, we advise in accordance with the recent recommendations of the European Centre for Diseases Prevention and Control regarding the screening regimen for carbapenem-resistant A. baumanni [30], to screen all patients, transferred from foreign countries with high prevalence rates MDR A. baumannii.