This was a retrospective, multicenter, and non-interventional epidemiological cohort study conducted between 2007 and 2009 in the ICU of nine hospitals in France, Switzerland, and Belgium. Patients with confirmed hospital-acquired pneumonia (HAP) and VAP caused by P. aeruginosa were analyzed for a period of 30 days from the time of diagnosis of pneumonia. The following ethics committees approved the study: Basel Ethikkommission beider Basel EKBB, Geneva Commission d’Éthique du Département de Médecine, Hôpitaux Universitaires de Gèneve, Lausanne Commission d’Éthique de la Recherche Clinique de la Faculté de Biologie et Médecine, le Comité de Protection des Personnes, Ile de France-VI, Brussels Université Catholique de Louvain, Faculté de Médecine, Commission d’Éthique Biomédicale Hospitalo-Facultaire. Because it was a retrospective epidemiological study, no informed consent was needed. Written informed consent was obtained in only 14 patients with serotype O11 HAP or VAP when they were included in a phase IIa study 19 .

Included patients were at least 18 years old, required intensive care management with HAP and VAP, were expected to survive longer than 72 hours, and fulfilled one of the following two criteria 20 :

1. Confirmed microbiological diagnosis of HAP and VAP caused by P. aeruginosa isolated from lower respiratory tract specimen with significant threshold of at least 1 × 10⁴ colony-forming units (CFU)/mL for bronchoalveolar lavage (BAL) 21 or 1 × 10³ CFU/mL for protected mini-lavage (mini-BAL) 22 , or below if under treatment with antibiotics, and presence of a new or progressing pulmonary infiltrate, plus one of the following three criteria: (a) fever of greater than 38°C, (b) white blood cell count of greater than 10,000/mm³, or (c) purulent sputum

Or

2. Confirmed microbiological diagnosis of HAP and VAP caused by P. aeruginosa isolated from endotracheal aspirate (ETA) with at least 1 × 10⁶ CFU/mL and a modified clinical pulmonary infection score (MCPIS) of higher than 6 points 23 .

Patients with nosocomial pneumonia caused by multiple bacteria were included in the study. Multiple bacteria were defined as more than one pathogen reaching significant threshold isolated from lower respiratory tract specimen.

During the same period of time and at the same centers, a phase IIa study which included 14 patients with O11 HAP or VAP was performed 19 . To reflect the correct prevalence of O11 serotype during the observation period, these patients were included in the prevalence calculation. They were, however, excluded from the clinical outcome analysis as they were treated with a combination of standard antibiotic treatment plus an adjunctive immunotherapy 19 . Exclusion criteria were applicable to only the 14 patients who participated in the phase IIa study. They included use of any investigational drug within 30 days prior to study commencement or during the study; patients with a known complement deficiency associated with systemic lupus erythematosus, paroxysmal nocturnal hemoglobinuria, hereditary angioedema, membranoproliferative glomerulonephritis, collagen vascular disease, autoimmune hepatitis, primary biliary cirrhosis, scleroderma, or recurrent Neisserial infections; or patients with confirmed HIV infection. Transplant patients and/or patients treated with systemic immuno-suppressive drugs (except prednisone or prednisolone), patients with a known liver function deficiency, and neutropenic patients (absolute neutrophil count of less than 1,000 cells/μL) were also excluded 19 .

Serotypes of P. aeruginosa isolated from lower respiratory tract specimens were determined by agglutination using serotype-specific antisera (Bio-Rad, Marnes-la Coquette, France) in accordance with the instructions of the manufacturer.

Patient clinical characteristics, Acute Physiology and Chronic Health Evaluation II (APACHE II) score 24 , sequential organ failure assessment (SOFA) score, MCPIS, risk factors, microbiological assessments (including serotyping) in ETA or BAL/mini-BAL, blood culture, chest radiographies, and laboratory parameters were assessed at inclusion. Scores, laboratory parameters, chest radiographies, and body temperature were collected as frequently as possible from day 1 to day 15, at days 21 and 30, or end of study (EOS) until patient died or was lost for follow-up.

Data collected at day 30 or EOS, whichever date was earlier, included clinical outcome, scores, laboratory parameters, chest radiographies, body temperature, discharge from the ICU, and discharge from the hospital.

Clinical outcome was defined as follows:

● Continuation: clinical signs and symptoms of pneumonia present during the whole assessment period until day 30 (or EOS)

● Resolution: complete resolution of pneumonia signs and symptoms present at the time of enrollment, no new symptoms or complications attributable to the pneumonia, no recurrence of pneumonia, and alive, until day 30 (or EOS)

● Recurrence: return of all clinical signs and symptoms of HAP/VAP, including infiltrates after initial clinical resolution

● Death: assessment period until day 30 (or EOS). Actual mortality was compared with predicted mortality according to APACHE II score 24 .

Microbiological outcome was defined as follows:

● Continuation: positive culture of P. aeruginosa in ETA or (mini)-BAL over the whole assessment period until day 30 (or E OS) associated with persisting clinical signs and symptoms of pneumonia

● Resolution: baseline isolate not present in repeated culture from original infection site

● Recurrence: isolation of P. aeruginosa from a culture taken after the resolution of pneumonia

● Colonization: positive culture of P. aeruginosa in ETA or (mini)-BAL over the whole assessment period until day 30 (or EOS) without any clinical pneumonia signs and symptoms.

Descriptive statistics were used for patient characteristics and outcome variables. Categorical variables were expressed as percentage, and continuous variables as mean ± standard deviation or as median and 25% to 75% interquartile range according to data distribution. Clinical and microbiological outcomes were considered binary endpoints and these were analyzed by using Bayesian logistic regression models. Logistic regression was used for binary outcomes and model based P values, using likelihood ratio tests. Time-to-event endpoints (survival and clinical and microbiological resolution) were analyzed by using Kaplan-Meier and Cox proportional hazards regression models. Analysis was done without adjustment for risk factors, with adjustment for different serotypes, and with adjustment for risk factors (age, APACHE II score, CPIS, SOFA score, and adequacy of antibiotic treatment). Analyses of data were performed by using R 2.10.1 and WinBUGS 1.4.3.

Patients’ flow chart is shown in Figure 1. Of 143 patients, 123 patients were included in the prevalence analysis and 129 in the overall clinical outcome analysis. Among 123 patients with serotype available, 114 patients had VAP (93%) and 9 had HAP (7%). Clinical and demographic data are shown in Table 1. At pneumonia onset, median length of hospital stay was 13 days, median length of ICU stay was 8 days, and median duration of mechanical ventilation was 8 days.

Data are expressed as median and 25%-75% interquartile. Nosocomial pneumonia (NP) includes hospital-acquired pneumonia and ventilator-associated pneumonia (VAP). APACHE II, Acute Physiology and Chronic Health Evaluation II; ICU, intensive care unit; MCPIS, modified clinical pulmonary infection score; NT, not typeable; PL5, sum of all serotypes with a prevalence of less than 5%; SOFA, sequential organ failure assessment.

The rates of prevalence of serotypes were, respectively, O6 (29%), O11 (23%), O10 (10%), O2 (9%), and O1 (8%). The sum of all serotypes with a prevalence of less than 5% (PL5) was 13%, and 8% of all patients had a P. aeruginosa infection classified as not typeable (NT) (Table 2).

Serotypes with a prevalence of more than 5% are expressed in boldface. NT, not typeable.

The prevalence of P. aeruginosa serotypes in French and Swiss investigator centers is shown in Figure 2. The distribution of P. aeroginosa serotypes by each investigator site is shown in Table 3.

Investigator sites with more than 10 patients were analyzed. Percentages of serotype are expressed in boldface. NT, not typeable; PL5, sum of all serotypes with a prevalence of less than 5%.

On day 30, across all serotypes, a 19% mortality, a 70% clinical resolution, an 11% clinical continuation, and a 5% clinical recurrence were recorded. Mortality was associated with age (P = 0.012) and high APACHE II score (P = 0.003). Clinical resolution was not associated with any of the risk factors—APACHE II score, CPIS score, SOFA score, adequate antibiotic treatment and age—or interaction of risk factors. Clinical continuation was associated with high APACHE II score (P <0.001).

Time to death was associated with age (P = 0.015) and high APACHE II score (P = 0.010). Time to event analysis of clinical resolution was associated with high APACHE II score (P = 0.008).

Among 102 patients with microbiological information available, a 41% microbiological resolution, a 25% microbiological continuation, a 23% colonization, and an 11% microbiological recurrence were recorded. Time to microbiological resolution was associated with APACHE II score (P <0.001) and higher MCPIS (P <0.007).

Among 109 patients (Figure 1), 15% of patients had nosocomial pneumonia caused by multiple bacteria, which were evenly distributed among the different serotype groups. An 18% mortality and a 69% clinical resolution were recorded. The highest mortality rate was observed in patients with serotype O1 infections (40%), and the lowest in patients with serotype O2 infections (0%) (Figure 3). Actual mortality and predicted mortality by APACHE II score were similar in patients infected by serotypes O1, O10, and O11, whereas actual mortality was lower than predicted mortality in patients infected by serotypes PL5, O6, NT, and O2 (Figure 3).

As shown in Figure 4, the highest clinical resolution rate was found in patients with serotype O2 (82%) and the lowest in patients with serotype O1 (40%). The serotypes O1, O10, O11, and NT had lower rates of clinical resolution (40%, 58%, 57%, and 60%) and higher rates of continuation (20%, 17%, 21%, and 30%), whereas the serotypes O6 and the PL5 group had continuation rates below 10%. The highest rates of recurrence were found in patients with serotype O1, whereas none of the 11 patients with serotype O10 and PL5 relapsed.

Mean time to death across all groups with nosocomial pneumonia was 11.4 days (Table 4). Overall time to resolution was 12.1 days, the shortest being for the patients with O2 infections (10 days) and longest being for the patients with O11 infections (13.4 days) (Table 4).

NT, non typeable; PL5, sum of all serotypes with a prevalence of less than 5%; SD, standard deviation.

While the prevalence of P. aeruginosa serotype varies from one hospital to another and from one country to another, O6 and O11 are often the most prevalent serotypes reported in previous studies 10 14 25 26 27 28 29 30 . It should be pointed out that the prevalence of P. aeruginosa serotypes reported in previous studies was obtained either from multiple infectious sites or from a single hospital or country. To the best of our knowledge, our study is the first assessing the prevalence of P. aeruginosa serotypes in nosocomial pneumonia from different hospitals of different countries.

Although the prevalence of serotypes is different among the countries and investigator sites (Figure 2 and Table 3), the overall rates of prevalence of the most common P. aeruginosa serotypes observed in patients with nosocomial pneumonia were O6 and O11. After exclusion of the largest enrollment site from the analysis, O6 and O11 remain the most prevalent serotypes, suggesting that investigator site effect does not influence interpretation of the finding. This result is comparable with those previously reported in the literature 10 15 25 26 28 . In approximately 8% of cases, the standard method failed to detect serotype because of self-agglutinating or non-agglutinating strains and the samples were classified as “not typeable”. This proportion is less than those previously reported 11 18 27 .

Age and APACHE II score were found to be the predictive risk factors for nosocomial pneumonia caused by P. aeruginosa. This result indicates that mortality and clinical and microbiological resolutions were strongly correlated with severity of the patients at the initial phase of pneumonia 31 32 33 .

The relationship between virulence of P. aeruginosa and different serotypes has been investigated in a few studies 12 13 14 . It has been shown that clinical strains lacking B-band O antigen increase TTSS and virulence 34 . On the other hand, it was found that strains secreting ExoU were frequently serotyped as O11 12 15 . ExoU is one of the toxins secreted by TTSS and contributes to epithelial cell toxicity, lung injury, and sepsis in infected animals 35 36 . Exo S and Exo T disrupt the host cell actin cytoskeleton, block phagocytosis, and are associated with mortality in animal models 37 . Furthermore, TTSS is associated with persistent VAP caused by P. aeruginosa and poor clinical outcome 38 39 . Faure et al. reported that, among 13 P. aeruginosa O1 strains, 7 secreted Exo S 12 . Recently, Le Berre et al. showed that the O11 serotype, elastase production, and TTSS were associated with increased lung injury in a murine model of pneumonia 15 . They found that serotype O11 strains were significantly more virulent than non-typeable strains and serotype O6; O11 strains were associated with a positive (ExoU) phenotype, whereas O6 strains were associated with negative ExoU phenotype 12 15 . These results suggest that clinical outcome of patients with pneumonia caused by P. aeruginosa could be related to serotypes.

This article reports on the first ever study to assess the clinical outcome of nosocomial pneumonia caused by a range of different serotypes of P. aeruginosa in critically ill patients under standard ICU management. Interestingly, the results show that the mortality rates for serotypes O1, O10, and O11 were in line with the expected mortality rates estimated by the severity of the patients (APACHE II score). Infections with serotypes O6, O2, and NT serotypes, however, had lower mortality than expected. There was a trend toward poorer outcome of patients with pneumonia caused by serotypes O1.

By cross-referencing prevalence and mortality rates, one can see that, while the O6 serotype is the most prevalent (29%), it has lower levels of continuation (8%) and death (17%) and higher resolution rates (75%) than the second most prevalent serotype O11 (prevalence 23%, continuation 21%, death rate 21%, and resolution 57%). This finding could be explained by the absence of ExoU with serotype O6 and presence of ExoU with serotype O11 as previously reported 15 . Consequently, compared with O6, O11 may be a more clinically relevant target for therapy in terms of patient outcomes 19 . Further study with larger groups is required in order to draw firm conclusions.

First, the prevalence obtained from our study may not be applied globally, because of an absence of data from additional countries known for their high prevalence of pneumonia caused by P. aeruginosa, such as Spain, Italy, Greece, the US, Mexico, Canada, and Japan. Serotypes may also vary from hospital to hospital within a given country and within the same hospital when assessed at different times. The most prevalent serotypes obtained in our study are, however, comparable to those previously reported in the literature. Second, the potential relationship between serotypes, virulence factors, and antibiotic resistance was not studied. Patients hospitalized in the same ICU could have a P. aeruginosa clone with the same virulence. Moreover, P. aeruginosa strains with the same serotype could be different clones exhibiting different virulence. Third, interpreting the results on clinical outcome is limited due to the small size of the individual serotype groups. However, these preliminary data provide relevant information for further investigations.

Further larger clinical study combining clinical outcome with distribution of virulence factors by serotypes and identification of specific clones by pulsed-field gel electrophoresis is needed to enhance the implication of these clinical findings 40 .