A total of 49 A. baumannii isolates from sputum were obtained between August and November of 2013 at Keimyung University Dongsan Medical Center, Daegu, Korea. All isolates were collected from patients newly infected or colonized by A. baumannii. The isolates were identified using conventional biochemical methods and 16S rRNA sequencing. A. baumannii ATCC 19606 (ATCC, Manassas, VA, USA) was used as a positive control in the biofilm formation assay, and Escherichia coli DH-5α (ATCC) was used as a negative control. A. baumannii ATCC 19606, E. coli DH-5α, and the 49 clinical isolates of A. baumannii were maintained on Muller-Hinton agar at 35°C.

Antibiotic susceptibility testing was conducted using a VITEK II system with NH cards (bioMérieux, Marcy-l’Étoile, France), following the methodology and breakpoints defined by the Clinical and Laboratory Standards Institute.

The following patient data were collected: age, sex, date of admission, date of discharge, underlying conditions, invasive procedures, antibiotic usage in the previous 1 month, infection or colonization, hospital stay before and after isolation, therapy, and 30-day mortality. Pneumonia was diagnosed according to criteria developed by the Centers for Disease Control and Prevention (United States) [10].

We defined colonization duration as the time from the day of the first A. baumannii-positive culture to the first day when A. baumannii was not detected in culture. Persistent colonization was the sustained detection of A. baumannii until the final culture report. In addition, we required the final culture study to have been conducted within 1 week of the day of discharge or death. Clear-up meant that A. baumannii was no longer reported by any subsequent culture tests. This study was approved by the Institutional Review Board of Keimyung University Dongsan Medical Center (IRB 2013-11-040-004).

Biofilm formation was assayed by crystal violet staining, as described previously [11]. Fresh bacterial suspensions were prepared from overnight cultures, and each were adjusted to an optical density (OD₆₀₀) of 0.1. The bacterial suspensions (100 μL) were inoculated in individual wells of a 96-well plate and incubated at 35°C for 24 hours. After an overnight incubation, the plates were gently washed twice with 200 μL phosphate-buffered saline (PBS), air-dried, and stained with 0.1% crystal violet (100 μL) for 15 minutes at room temperature. Plates were gently washed twice with PBS, the stains were solubilized with 99% ethanol, and the OD₅₇₀ of the supernatant in each well was measured using a Victor 3 microplate reader (PerkinElmer, Waltham, MA, USA). Biofilm-producing isolates were defined as those for which the optical density of solubilized crystal violet in 99% ethanol was higher than the average optical density of solubilized crystal violet for A. baumannii ATCC 19606. Average optical density of solubilized crystal violet for A. baumannii ATCC 19606 was calculated using measurements of the positive control well on each plate. All experiments were performed in triplicate. Each plate included the following controls: media alone, A. baumannii ATCC 19606 (positive control), and E. coli DH-5α (negative control).

Data management and statistical analyses were performed with SPSS software version 21.0 (IBM Co., Armonk, NY, USA). All data were first subjected to bivariate analysis. Categorical variables were compared using the chi-square test or Fisher exact test, and continuous variables were compared using the Mann-Whitney U test. Conditions contributing to colonization duration were analyzed using chi-square univariate analysis; those conditions that were significant were included in a subsequent multiple logistic regression analysis to calculate confidence intervals. All tests of significance were two-tailed. We considered tests significant when they had p values of 0.05 or below.

The ability of A. baumannii ATCC 19606 to form a biofilm is established [12]. Of the 49 isolates examined, 24 (48%) exhibited enhanced biofilm formation capacity relative to a standard A. baumannii strain (Fig. 1).

Antibiotic resistances were similar between the biofilm-producer group and the non-producer group. All isolates were carbapenem resistant A. baumannii. Thirty-eight isolates (77%) were collected from patients in the ICU and a total of 47 patients (95%) were exposed to antibiotics within the previous month. Resistance to aminoglycosides and tigecycline seemed to be higher in the biofilm non-producer group than in the biofilm producer group, with no statistical significance (Table 1).

The average patient age was lower for the biofilm producer group (66.5 vs. 75, p = 0.017). There was no statistical difference in the underlying diseases of patients harboring biofilm producers versus those harboring biofilm non-producers. Twenty-seven patients (55%) had an endotracheal tube and 17 (34%) were on mechanical ventilation. No underlying patient disease was more common to the biofilm producer group (Table 2).

Thirty-seven isolates (75%) had persistently colonized their host patient’s respiratory tract. The median duration of colonization was longer for biofilm-producing isolates than it was for biofilm non-producing isolates. Simultaneous colonization with other bacteria was more common for biofilm-producing isolates. The most prevalent co-colonizing bacteria were Staphylococcus aureus. The median timespan from admission to acquisition seemed to be shorter for the biofilm-producing group; however, this difference was not statistically significant (Table 3).

We conducted a multivariate analysis to confirm the relationship of biofilm formation and colonization duration. In the multivariate analysis, isolates with colonization durations of more than 2 weeks are significantly associated with biofilm and ICU stays of more than 2 weeks (Table 4).