|Home | About | Journals | Submit | Contact Us | Français|
Pseudomonas aeruginosa biofilms exhibit increased antimicrobial resistance compared with planktonic isolates and are implicated in the pathogenesis of both acute and chronic lung infections. Whilst antibiotic choices for both infections are based on planktonic antibiotic susceptibility results, differences in biofilm-forming ability between the two diseases have not previously been explored. The aim of this study was to compare differences in biofilm formation and antibiotic resistance of P. aeruginosa isolated from intubated patients and from patients with chronic pulmonary disease associated with cystic fibrosis (CF). The temporal evolution of antibiotic resistance in clonal P. aeruginosa strains isolated from CF patients during periods of chronic infection and acute pulmonary exacerbation was also evaluated. Biofilm formation and biofilm antibiotic susceptibilities were determined using a modified microtitre plate assay and were compared with antibiotic susceptibility results obtained using traditional planktonic culture. Clonality was confirmed using random amplified polymorphic DNA polymerase chain reaction (RAPD-PCR) analysis. Pseudomonas aeruginosa isolates collected from intubated patients produced substantially more biofilms compared with CF isolates. There was considerable heterogeneity in biofilm-forming ability among the CF isolates and this was unrelated to pulmonary status. Biofilm antibiotic resistance developed rapidly among clonal CF isolates over time, whilst traditional antibiotic resistance determined using planktonic cultures remained stable. There was a significant positive correlation between imipenem/cilastatin and ceftazidime resistance and biofilm-forming ability. The variability in biofilm-forming ability in P. aeruginosa and the rapid evolution of biofilm resistance may require consideration when choosing antibiotic therapy for newly intubated patients and CF patients.
Pseudomonas aeruginosa is the leading cause of morbidity and mortality in patients with ventilator-associated pneumonia (VAP) and cystic fibrosis (CF). Whilst VAP and CF are notably different pulmonary disease processes, biofilm formation is a unifying process in the pathogenesis both of acute lung infections such as VAP and chronic lung infections such as CF.
Biofilms are structured communities of bacterial cells enclosed in an extracellular secreted polymeric matrix consisting of secreted proteins, exopolysaccharides and nucleic acids that can adhere both to abiotic and living surfaces, serving as a permanent source of infection . Biofilms have been shown to be up to 1000 times more resistant to antibiotics than planktonic, or free-swimming, cells of the same isolate . This is due to a number of mechanisms, including the exopolysaccharide matrix acting as a physical barrier to antibiotic penetration and the creation of an antibiotic gradient throughout the biofilm. Exposure of cells within the biofilm to sublethal concentrations of antibiotics further promotes antimicrobial resistance in cells that may then detach from the biofilm and disseminate infection elsewhere. Oxygen and nutrient depletion may cause the bacteria to enter a non-growing or stationary phase, which increases resistance to antibiotics such as β-lactams . In addition, biofilm-specific antimicrobial resistance genes not expressed during the planktonic phase have been shown to increase resistance of cells in these sessile communities .
There are notable differences in the pathophysiology and presentation of VAP and CF. In mechanically ventilated (MV) patients, injury due to insertion of the endotracheal tube (ETT) and the absence of host defences facilitate the entry of bacteria in the tracheal mucosa. Bacterial colonisation and biofilm formation can rapidly occur on the inner lumen of the ETT as early as 12 h after intubation [5–7]. The biofilm can then be dislodged from the ETT with suctioning and disseminated towards the lower respiratory tract, potentially leading to VAP. VAP with P. aeruginosa is an invasive and rapidly progressing infection characterised by acute leukocytosis, fever and increased need for ventilator support. In contrast, initial infection with P. aeruginosa in CF patients typically occurs in childhood and infection is lifelong and persistent. The thick mucus layer overlaying CF airway epithelial cells is particularly amendable to the formation of biofilms. With the establishment of chronic infection, P. aeruginosa may convert to a mucoid phenotype (i.e. hyperproduction of alginate), resulting in biofilms that are virtually impossible to clear by conventional antimicrobial therapy. Whilst chronic colonisation with P. aeruginosa biofilms in CF patients is typically subclinical and silent, acute exacerbations can lead to respiratory decompensation, likely due to release of planktonic bacteria from biofilm colonies . The continual persistent infection, in addition to host defences, leads to chronic airway inflammation and eventual airway destruction. In contrast to P. aeruginosa VAP, pulmonary exacerbations in CF patients are episodic with only mild elevations in temperature and white blood cell (WBC) count.
In this study, the biofilm-forming ability and associated antibiotic resistance of P. aeruginosa recovered from two distinct patient populations, namely MV patients newly colonised with multidrug-resistant (MDR) P. aeruginosa and CF patients during periods of chronic infection and episodes of acute pulmonary exacerbation, were compared. Serial P. aeruginosa isolates from CF patients were examined to evaluate the evolution of antimicrobial resistance both in planktonic and biofilm cultures over time.
Daily endotracheal aspirate (ETA) surveillance cultures were performed on patients ≥18 years of age who were intubated for ≥48 h in the Intensive Care Unit (ICU) irrespective of signs or symptoms of lung infection . Patients with known infection or risk factors for colonisation with P. aeruginosa (CF, bronchiectasis, chronic tracheostomy, previous intubation during the current hospitalisation or a previous culture positive for P. aeruginosa) were excluded. Six patients were found to be newly colonised with a MDR P. aeruginosa isolate. These six isolates where chosen for evaluation of biofilm-forming ability and associated antimicrobial resistance. Clinical characteristics were noted at the time of positive culture. VAP was defined as any lower respiratory infection that developed after 2 days of mechanical ventilation. Criteria for VAP were presence or progression of a new infiltrate identified by chest radiography plus two of the following three clinical criteria: (i) fever >38 °C; (ii) WBC count >12 000 cells/mm3 or <4000 cells/mm3; and (iii) purulent secretions and ≥106 colony-forming units (CFU)/mL P. aeruginosa from the ETA and/or ≥104 CFU/mL from a bronchoalveolar lavage .
To examine the evolution of antibiotic resistance over time in patients with CF, serial respiratory cultures were collected from three CF patients known to be chronically infected with P. aeruginosa over a 3-month period during their regularly scheduled outpatient clinic appointment and when hospitalised for treatment of acute pulmonary exacerbation as part of an ongoing prospective cohort study. Five isolates were collected from Patient 1 on Days 1, 28, 29, 30 and 84, three samples were collected from Patient 2 on Days 1, 2 and 62 and two samples were collected from Patient 3 on Days 1 and 61. Clinical characteristics [i.e. age, sex, forced expiratory volume in 1 s (FEV1) and current antibiotics] were noted at the time of each sample. Patients were noted to be either chronically infected yet clinically stable or in acute pulmonary exacerbation requiring hospitalisation. The Institutional Review Board of University of California, San Francisco (San Francisco, CA) approved all protocols.
A DNA fingerprint of each isolate was generated by RAPD analysis to determine clonality of the P. aeruginosa strains. DNA was extracted from overnight Luria–Bertani (LB) cultures of each strain using the Promega Wizard® Genomic DNA Purification Kit (Promega, Madison, WI). RAPD primer 272 was used as described previously .
Tobramycin (Teva Pharmaceuticals, Irvine, CA), ceftazidime (GlaxoSmithKline, Research Triangle Park, NC), piperacillin/tazobactam (TZP) (Wyeth, Philadelphia, PA), imipenem/cilastatin (Merck and Co, Whitehouse Station, NJ) and levofloxacin (Ortho-McNeil-Janssen, Raritan, NJ) were used. Fresh stock solutions of each antibiotic were prepared in sterile water. Concentrations tested ranged from 0 µg/mL to 256 µg/mL for each drug.
Biofilms of each P. aeruginosa isolate were grown according to the method of O’Toole and Kolter . Overnight cultures of all isolates were grown to stationary phase and were diluted to an optical density at 650 nm (OD650) of 0.1 in fresh LB medium. Then, 200 µL of the diluted culture was used to inoculate 8 wells per isolate of a clear 96-well microtitre plate (Nunc TSP System; Nunc International, Rochester, NY). Peg lids (TSP lids; Nunc International) were used to cover the plates and plates were statically incubated at 37 °C for 24 h. The biofilms formed on the immersed pegs were stained with 0.1% w/v crystal violet (125 µL) for 10 min, rinsed and allowed to dry for several hours. To solubilise absorbed crystal violet, pegs with stained biofilms were incubated in 70% ethanol (200 µL/well) for an additional 10 min. The OD650 was measured on a microtitre plate colorimeter (Molecular Devices, Sunnyvale, CA). Mean readings from eight wells were calculated to determine the biofilm-forming ability of each strain. Pseudomonas aeruginosa strains PA01 and PA103 were included as control strains, as PA01 has been shown to form robust biofilms while PA103 forms weak biofilms.
The minimum inhibitory concentration (MIC) of planktonic P. aeruginosa was determined for each antibiotic using standard 96-well microbroth dilution plates (Dade Behring, West Sacramento, CA). Results were interpreted according to Clinical and Laboratory Standards Institute (CLSI) guidelines .
The biofilm inhibitory concentration (BIC), i.e. the antibiotic concentration needed to inhibit 90% of biofilm viability, was determined using a modified version of the method of Moskowitz et al. . In brief, biofilms were allowed to form on peg lids as described above. Following biofilm formation, peg lids were transferred to a fresh microtitre plate containing 0–256 µg/mL of antibiotic in a total volume of 100 µL/well. Eight replicate wells were used for each concentration tested for each strain. Plates were re-incubated statically at 37 °C for 24 h (antibiotic challenge). Following incubation in antibiotic, peg lids were directly transferred to another 96-well plate containing fresh LB medium prior to centrifugation at 2500 × g for 35 min to remove biofilm cells from the peg lids (biofilm recovery). The peg lid was replaced with a standard lid and residual biofilm viability was measured at each antibiotic concentration as the difference in mean optical density at 600 nm before and after incubation for 6 h at 37 °C (OD600 at 6 h – OD600 at 0 h). The mean biofilm viability of eight wells per antibiotic concentration tested was calculated.
Regression analyses were performed in the R statistical environment to determine whether a statistically significant correlation between biofilm formation and biofilm resistance for each of the antibiotics tested existed. P-values of ≤0.05 were considered statistically significant. Regression plots were constructed using Microsoft Excel (Microsoft Corp., Redmond, WA).
The characteristics of the six MV patients and three CF patients are shown in Table 1. Of the MV patients, the mean age at study inclusion was 63 years (range 50–79 years). The mean time from start of mechanical ventilation to colonisation with MDR P. aeruginosa was 17.7 days (range 2–26 days). Four patients died. Carbapenems (e.g. imipenem and meropenem) and linezolid were the most common antibiotics received at the time of colonisation. One patient (Patient 3) was not receiving any antibiotics at the time of colonisation.
Of the three CF patients, the most common antibiotics utilised during episodes of acute pulmonary exacerbation included a β-lactam and either an aminoglycoside or a fluoroquinolone.
RAPD analysis demonstrated that each of the P. aeruginosa strains isolated from the MV patients was unique (Fig. 1). Sequential isolates from each of the CF patients were clonal, suggesting that the temporal increase in antimicrobial resistance observed in these strains was due to evolution rather than acquisition of a novel P. aeruginosa strain.
Isolates from MV patients exhibited consistently strong biofilm-forming ability compared with strains from CF patients and control strains PA01 and PA103 (Fig. 2). Biofilm formation was comparatively weak for the CF isolates. There was also greater variability in biofilm formation among the CF isolates. There was no correlation between biofilm formation and pulmonary status (e.g. chronic stability versus acute exacerbation).
MICs for isolates from the six MV patients and three CF patients are presented in Table 2. Five (83%) of the six MV isolates were resistant to ceftazidime, imipenem/cilastatin and levofloxacin according to CLSI breakpoints. All MV isolates were susceptible to tobramycin. TZP exhibited the most variability in susceptibility, with MICs ranging from 8 µg/mL to 128 µg/mL.
In contrast, the MICs of the CF isolates were more susceptible compared with the MV isolates. Ceftazidime, imipenem and levofloxacin susceptibilities remained stable over time in all three CF patients, with either no increase in MIC (ceftazidime and levofloxacin) or a one-fold increase (imipenem, CF1). Interestingly, the Day 30 isolate from Patient 1 exhibited a two-fold decrease in levofloxacin susceptibility. Resistance evolution developed over time for TZP and tobramycin. There was a two-fold increase in the MIC of TZP over time for Patients 1 and 3. Resistance development was greatest with tobramycin and was observed in all three patients over time.
Among the MV patient isolates, the BIC was found to be greater than the MIC for all antibiotics except one (i.e. the BIC for MV5 tested against ceftazidime remained stable) (Table 2). The β-lactam antibiotics exhibited the greatest increase in antibiotic resistance (BICs ≥ 256 µg/mL for five of the six isolates). Levofloxacin BICs were found to be two to five doubling dilutions higher than the corresponding MIC; tobramycin BICs were found to be three to five doubling dilutions higher than the corresponding MIC.
Among the CF isolates, the increase in BIC over MIC was modest compared with the MV isolates. Among the β-lactam antibiotics, the greatest increases were observed with TZP. Levofloxacin and tobramycin exhibited the greatest variability in BICs over time. Tobramycin BICs were found to range from 0.25× MIC (CF1, Day 29 isolate) to ≥8× MIC. This result was confirmed with repeat testing. A decrease in BIC was also observed for levofloxacin (CF2, Day 62). The temporal evolution of biofilm resistance was most pronounced with isolates tested against levofloxacin and tobramycin.
Significant positive correlations between BIC and biofilm formation existed for imipenem (r = 0.83, P < 0.001), levofloxacin (r = 0.57, P < 0.021) and ceftazidime (r = 0.62, P < 0.01), suggesting that resistance towards these antimicrobials strongly correlated with biofilm formation (Fig. 3). No significant correlation was observed between biofilm formation and tobramycin (r = 0.41, P < 0.107). A negative correlation was observed between biofilm formation and TZP, although this was not statistically significant (r = 0.32, P < 0.194).
Biofilms have been implicated in virtually every human infection and are particularly recalcitrant to antibiotic compounds and can persist despite sustained host defences. Biofilm formation is an important component of pulmonary colonisation and infection due to P. aeruginosa both in acute lung infections such as VAP and in chronic lung infections such as CF. Understanding bacterial physiology and the mechanisms by which P. aeruginosa protects itself from lethal concentrations of antibiotics is fundamental to designing interventions against resistant strains. This study compared the biofilm-forming ability of P. aeruginosa newly colonising the airway of MV patients and patients with longstanding CF during periods of chronic stability and acute pulmonary exacerbation and its associated effect on antibiotic susceptibility.
Biofilm-forming ability was greatest among P. aeruginosa biofilms from newly colonised MV patients. Enhanced biofilm formation by isolates from ETAs suggests that these strains may have originated form P. aeruginosa biofilms formed on the inner lumen of the ETT. Biofilm formation on the ETT is facilitated by multiple factors, including the polyvinylchloride plastic surface of the ETT, the lack of host defence mechanisms in the inner lumen and the accumulation of tracheobronchial secretions. Biofilms can then be dislodged from the ETT during routine suctioning and disseminate into the lower respiratory tract, contributing to the development of VAP. Whilst biofilm formation of P. aeruginosa isolates colonising the ETT was high among the MV patients, only 50% of patients went on to develop VAP. Adair et al.  showed that while 100% of patients (20/20) with VAP had ETTs covered with biofilm, 30% of patients (6/20) without VAP also had ETTs colonised with biofilms. Thus, while ETT colonisation is an important contributor to VAP, it is only one of many notable risk factors, including the severity of the patient’s underlying disease, prior surgery and prior exposure to antibiotics [15,16].
In contrast, biofilm formation was lowest among the CF isolates and was highly variable among sequential clonal isolates. There was no correlation between biofilm formation and pulmonary status (e.g. chronic infection versus acute pulmonary exacerbation). The variability in biofilm formation among P. aeruginosa CF isolates with similar genotypic profiles is supported by others . However, biofilm formation in relation to pulmonary status in sequential clonal isolates in CF patients has never been investigated. Differences in host factors and antibiotic selection pressure may account for the variability in biofilm formation and pulmonary status in CF patients. Impaired mucociliary clearance, the thick mucus layer overlying lung epithelium, and the presence of a microaerophilic or anaerobic environment are all factors that encourage the formation of P. aeruginosa biofilms in CF . The reason for the heterogeneity in biofilm formation between MV and CF patients is likely multifactorial. Biofilms formed on living tissue have been shown to be morphologically different from biofilms formed on abiotic surfaces such as ETTs. Other factors include differences in host defences, prior antibiotic exposure and the differential expression of bacterial virulence factors .
Comparison of MICs and BICs demonstrated a substantial difference in antibiotic resistance among the MV isolates. The greatest increase in biofilm antibiotic resistance occurred with the β-lactam antibiotics, with the majority of isolates exhibiting BICs ≥ 256 µg/mL. In contrast, the differences between the BIC and MIC for CF isolates were not as pronounced. At subinhibitory imipenem concentrations, imipenem has been shown to induce ampC production, increase biofilm volume and increase alginate production in P. aeruginosa PA01 biofilms and to induce biofilm formation in MDR Acinetobacter baumannii [20,21]. Receipt of subtherapeutic concentrations of imipenem in MV patients likely contributed to the increased biofilm formation and potentially increased ampC expression. This increased ampC expression, in turn, may explain the increased ceftazidime and TZP resistance observed in this study. In contrast, the majority of CF isolates were susceptible to imipenem and none of the patients received imipenem for treatment of their acute pulmonary exacerbations, leading to potentially less induction of biofilm formation and ampC expression. Tobramycin has also been shown to induce biofilm formation at subinhibitory concentrations in PA01. However, biofilm induction occurred in only approximately one-half of CF clinical isolates in a recent study and was independent of tobramycin susceptibility . This may explain the variability in biofilm-forming ability and BIC observed among the CF isolates in this study. It was also observed that resistance evolution in biofilms occurred rapidly, whilst resistance evolution in the planktonic mode of lifestyle remained relatively unchanged over time.
Regression analysis of biofilm formation with antimicrobial resistance demonstrated statistically significant positive correlations for three of the five antimicrobials tested. This suggests that physiological aspects specific to biofilms, such as exopolysaccharide production, biofilm-specific efflux pump expression or β-lactamase production, played a role in increased biofilm antimicrobial resistance. Interestingly, a negative correlation existed between biofilm formation and TZP resistance. Subinhibitory concentrations of TZP have been shown to lead to a decrease in bacterial adherence and biofilm formation among P. aeruginosa clinical isolates . In general, increased antimicrobial resistance and biofilm formation were associated, underlining the importance of this mode of lifestyle in clinical infections and when considering treatment options.
Because of the important role of biofilms in the pathophysiology of VAP and CF, strategies targeting the biofilm have been developed. These include selective decontamination of the ETT and use of antiseptic- or antimicrobial-impregnated ETTs for prevention of VAP. In the North American Silver-Coated Endotracheal Tube (NASCENT) study, Kollef et al.  randomised 2003 patients expected to require MV for ≥24 h to undergo intubation with either a silver-coated ETT or a conventional uncoated ETT. Compared with the uncoated ETT group, the silver-coated ETT group had a lower incidence of microbiologically confirmed VAP [37/766 (4.8%) vs. 56/743 (7.5%); P = 0.03] and a statistically significant delay in the onset of VAP. Other strategies studied include removal of established biofilms and mucus secretions in the ETT using mechanical devices such as the Mucus Shaver. The Mucus Shaver has been shown to be effective in removing biofilm from the inner lumen of the ETT in intubated sheep [25,26]. This technique is promising but further studies will be needed to determine whether removal of the biofilm is effective in preventing VAP. Whilst silver-coated ETTs are currently recommended by some groups, the Mucus Shaver cannot be recommended at this time in the absence of clinical data .
Inhaled antibiotics (e.g. tobramycin and colistin) have also been studied as a method to prevent biofilm formation in the ETT and as adjunctive therapy for the treatment of VAP [28–33]. Compared with systemic administration of antibiotics, inhaled antibiotics can achieve higher pulmonary concentrations, with levels far in excess of planktonic MICs [34,35]. Falagas et al.  conducted a meta-analysis of the effect of antibiotics administered via the respiratory tract for the prevention of ICU pneumonia. The incidence of VAP was reduced but there was no overall effect on mortality. Three recently published studies compared adjunctive inhaled colistin or colistimethate plus intravenous (i.v.) antibiotics with i.v. antibiotics alone for the treatment of VAP [29,30,37]. Using a matched retrospective case–control design, Kofteridis et al.  found no difference in eradication of Gram-negative organisms, including P. aeruginosa, in the treatment group compared with the control group [17/34 patients (50%) vs. 19/42 patients (45%); P = 0.679]. There was a trend towards improved clinical cure with inhaled plus i.v. colistin, but no statistically significant differences in either all-cause mortality or VAP-related mortality. Using a retrospective cohort analysis, Korbila et al.  found inhaled colistin to be independently associated with VAP cure but also no improvement in mortality. In a recent randomised controlled trial, Rattanaumpawan et al.  found that microbiological eradication was significantly improved in the treatment group compared with the control group (60.9% vs. 38.2%, P = 0.03), but again no difference in clinical outcome. Thus, inhaled antibiotics may have a favourable effect on microbiological outcome, but this technique has not yet been shown to translate into improved patient outcomes.
The most promising strategy is to dose antibiotics based on biofilm susceptibilities in lieu of planktonic susceptibilities. Keays et al.  retrospectively reviewed 110 CF patients who received antibiotic treatment for acute pulmonary exacerbation. Patients who received antibiotic combinations that inhibited bacteria (based on BIC testing) had a decrease in sputum bacterial density and hospital length of stay. Clinical trials examining the effectiveness of antibiotic dosing based on biofilm susceptibilities in CF are currently ongoing [39,40]. Whilst this strategy has not yet been studied for the treatment of VAP, this is an area of potential benefit that will require further investigation.
This study is associated with certain limitations. Previous investigators have suggested that use of more mature biofilms may be more representative of in vivo conditions in CF and exhibit increased antibiotic resistance compared with biofilms formed after only 24 h . However, ETAs were collected from MV patients on a daily basis, thus the biofilm formed on the ETT of these patients would likely be more representative of young biofilm. Use of a 24-h biofilm model allowed direct comparisons between the CF and MV isolates to be made. Owing to the small number of MV and CF patients examined, additional studies examining biofilm formation and resistance in larger cohorts of patients will be needed.
These results indicate that the effect of antibiotics on biofilm formation is highly variable and unpredictable, with some antibiotics inducing biofilm formation while other antibiotics inhibit biofilm formation in these two patient populations. This unpredictability in biofilm formation and BIC make it increasingly difficult for clinicians to choose the most active antibiotic, particularly as biofilm susceptibility testing is not routinely performed as part of clinical care. In addition, the rapid induction and evolution of biofilm resistance in clonal sequential CF isolates, whilst planktonic antibiotic resistance remains stable, is worrisome. Biofilm formation and the associated development of antimicrobial resistance is a multifactorial process dependent on the presence or absence of prosthetic material, host immune status and disease state. These are factors that will require further investigation in order to determine the optimal antibiotic therapy to successfully treat biofilm-mediated infections.
The authors thank Marshall Baek for technical assistance with this study.
This project was funded by grant no. KL2 RR024130 from the National Center for Research Resources (KY), the American Lung Association (SVL), NIH award AI075410 (SVL), NIH grants SCCOR HL 74005, HL 69809 and HL074005 (JW-K), and NIH award UO1 1AI075410 (SVL and KY).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Approval was obtained from the University of California, San Francisco (UCSF) Human Research Protection Program (approval nos. H53004-28671-05 and H1287-21213).