Risk Factors for a First Episode of Ventilator-Associated Pneumonia Caused by Stenotrophomonas Maltophilia

Background: The incidence of ventilator-associated pneumonia caused by Stenotrophomonas maltophilia (SM-VAP) is on the rise. This pathology is associated with increased morbidity and mortality in intensive care unit (ICU), notably due to intrinsic resistance and ineffective probabilistic antibiotic therapy. Our study aimed to determine the risk factors for a rst episode of SM-VAP in ICU. Methods: This single center retrospective study was conducted from 2010 to 2018 in the polyvalent ICU of Félix Guyon University Hospital in Reunion Island. All patients who developed ventilator-associated pneumonia (VAP) during their ICU stay were consecutively evaluated. Patients with a rst episode of SM-VAP were compared to those with a rst episode of VAP caused by another microorganism. Results: A total of 89 patients developed a rst episode of SM-VAP over the study period. In the group of patients with SM-VAP, infection was polymicrobial in 43.8% of cases and ICU mortality was 49.4%. After multivariate logistic regression analysis, the risk factors for a rst episode of SM-VAP were: chronic respiratory failure (Odds Ratio (OR): 4.212; 95% Condence Interval (CI): 1.776 – 9.989; p = 0.001), chronic renal failure (OR: 2.693; 95% CI: 1.356 – 5.352; p = 0.05), use of third-generation cephalosporins active against Pseudomonas aeruginosa (OR 2.862; 95% CI: 1.505 – 5.442; p = 0.001), and female sex (OR: 2.646; 95% CI: 1.458 – 4.808; p = 0.001). Conclusion: In our study, chronic respiratory failure, chronic renal failure, use of third-generation cephalosporins active against P. aeruginosa, and female sex were identied as risk factors for a rst episode of SM-VAP.


Background
Stenotrophomonas maltophilia (SM) is a ubiquitous bacterium mainly responsible for opportunistic and nosocomial infections. This strictly aerobic, non-fermenting gram-negative bacillus (NF-GNB) has its natural reservoirs in water, soil, and plants. While SM is found in health care facility water systems, it can also survive on inert surfaces by producing bio lm on medical material, which then acts as a vector of nosocomial infection (1). Until about 15 years ago, SM was considered to be of low pathogenicity, with the literature reporting numerous cases of colonization with no prognostic impact on infected subjects (2).
Studies have highlighted the intrinsic resistance of SM to several antibiotics -including extendedspectrum antibiotics, which are usually administered to intensive care unit (ICU) patients with serious pulmonary infection as part of probabilistic antibiotic therapy (3). The main mechanisms involved are altered membrane permeability, e ux pumps, and the presence of inducible chromosomal betalactamases (4).
According to epidemiological surveillance studies, SM is now the third most common infection-causing NF-GNB, after Pseudomonas aeruginosa and Acinetobacter baumannii (5,6). The prevalence of SM infections is steadily increasing (4), the most frequent being respiratory tract infections. Depending on the series, the incidence of ICU-acquired pneumonia caused by SM varies between 0.37 and 2.0% (7)(8)(9). A recent study conducted in our ICU found an incidence of ventilator-associated pneumonia caused by S. maltophilia (SM-VAP) of 1.4% (10).
Infections with SM are now considered to be particularly severe. Attributable mortality has been shown to vary between 12.0 and 60.0% for all types of infection (6,7,11) and between 41.2 and 50.0% for pneumonia (including ventilator-associated pneumonia (VAP)) (8,9,12,13). The excess mortality due to SM in patients with VAP could be explained by delayed initiation of effective antibiotic therapy.
Indeed,this microorganismis not speci cally covered by recommendations for the management of VAP (14,15). In the 2008 meta-analysis by Kuti et al., ineffective probabilistic antibiotic therapy was associated with increased mortality in patients with SM-VAP (16). Other series have since con rm this trend (17)(18)(19).
In view of the above, this study aimed to determine the risk factors for a rst episode of SM-VAP in ICU.

Methods
This single center retrospective study was conducted in the polyvalent ICU of Félix Guyon University Hospital in Reunion Island. All adult ICU patients who developed an episode of VAP between 1 January 2010 and 31 December 2018 were consecutively evaluated ( Figure 1). Data on rst episodes of VAP were retrospectively collected from the database of nosocomial infections of the french network ReaRaisin.
This study was approved by the Ethics Committee of the French Society of Infectious Disease and Tropical Medicine (CER-MIT 2021-0506) and was registered in the internal registry of treatment activities of Félix Guyon University Hospital (RCH-2021-0017). In accordance with the French legislation on noninterventional studies, the requirement for informed consent was waived owing to the retrospective nature of the study (20), this study agrees with the recommendations STROBE (The Strengthening the Reporting of Observational Studies in Epidemiology) (21).

De nitions
The diagnosis of pneumonia was con rmed in the presence of new or progressive radiological in ltrates associated with one of the following criteria: temperature > 38.5°C or < 36.0°C, white blood cell count >11G/L or <4G/L, purulent bronchial secretions, drop in oxygenation, and/or positive respiratory culture obtained by bronchoalveolar lavage, protected distal sampling, or endotracheal aspiration. Ventilator-associated pneumonia was de ned by the onset of pneumonia at least 48 hours after mechanical ventilation (MV) initiation (22)(23)(24).
Antibiograms were performed using the disc diffusion method and interpreted according to EUCAST (European Committee on Antimicrobial Susceptibility Testing) susceptibility breakpoints: 10 4 colony forming unit (CFU)/mL for bronchoalveolar lavage, 10 3 CFU/mL for protected distal sampling, and 10 5 CFU/mL for endotracheal aspiration (25).

Data collection
The following information was collected: -demographic data: age, sex.
-reason for admission to ICU.
-Sequential Organ Failure Assessment (SOFA) score on admission to ICU and on the day of VAP diagnosis.
-settings of MV.
-organ failure during ICU stay and on the day of VAP diagnosis.
-highest bilirubin level, lowest platelet count, lowest prothrombin level, and lowest PaO2/FiO2 ratio during ICU stay; PaO2/FiO2 ratio on the day of VAP diagnosis.
-use of organ support such as extracorporeal membrane oxygenation, renal replacement therapy, catecholamines.
-exposure data: prior corticosteroid therapy, prior rectal or respiratory colonization with SM, hospital stay in the last 6 months, type and duration of antibiotic therapy during ICU stay, parenteral nutrition during ICU stay, MV duration before VAP diagnosis.
-infection data: type of respiratory sample, type of infection (monomicrobial or polymicrobial), Clinical Pulmonary Infection Score (CPIS) on the day of VAP diagnosis, time to initiation of effective antibiotic therapy. Note that probabilistic antibiotic therapy was considered effective if at least one of the used molecules had in vitro activity against the isolated microorganism.
-prognosis: MV duration, tracheotomy, length of stay in ICU and in hospital, ICU and hospital mortality.

Statistical analyses
Results were expressed as number (percentage) for categorical variables and as median [25 th -75 th percentiles] for continuous variables. Categorical variables were compared using the chi-square test, the Kruskal-Wallis test, or Fisher's exact test, as appropriate. Continuous variables were compared using the nonparametric Mann-Whitney test.
Risk factors associated with a rst episode of SM-VAP with a p <0.05 in univariate analysis were entered into the multivariate logistic regression model. In cases of collinearity, only the most clinically relevant factors were entered in the multivariate model. Model calibration was assessed using the Hosmer-Lemeshow goodness-of-t test. The signi cance level was set at p<0.05. Statistical analyses were performed using SAS statistical software (8.2, Cary, NC, USA).

Study population
A total of 9,542 patients were admitted to the polyvalent ICU of Félix Guyon University Hospital between 1 January 2010 and 31 December 2018. Of these, 431 patients had an episode of VAP and were included in the analysis. Stenotrophomonas maltophilia was found in 20.6% (n = 89) of respiratory samples ( Figure  1).
The clinical and demographic characteristics of included patients are listed in Table 1.
Respiratory samples were obtained by protected distal sampling in 46.8% of cases, endotracheal aspiration in 34.5% of cases, and bronchoalveolar lavage in 18.7% of cases.
The prevalence of polymicrobial infection in patients infected with SM was 43.8%, with the most common co-infecting microorganism being P. aeruginosa ( Table 3). The prevalence of polymicrobial infection in patients infected with another microorganism was 34.5%. Identi ed SM strains had an identical rate of susceptibility of 84.2% to trimethoprim-sulfamethoxazole and to uoroquinolones. The rate of susceptibility to other antibiotics was lower, namely 61.8% for ticarcillin-clavulanate and 39.3% for ceftazidime. Probabilistic antibiotic therapy was effective against SM in 66.3% of cases. In the control group, probabilistic antibiotic therapy was effective against the identi ed microorganism in 94.2% of cases (p < 0.001).

Risk factors for SM-VAP in univariate analysis
The patient characteristics signi cantly associated with an increased risk of developing a rst episode of SM-VAP in univariate analysis were: female sex (p = 0.001), chronic respiratory disease (p = 0.013), chronic renal failure (p = 0.038), and malnourishment (p < 0.001) ( Table 1).

Discussion
To our knowledge, this study is the rst to evaluate the risk factors for a rst episode of SM-VAP. The following risk factors were identi ed: chronic respiratory disease, chronic renal failure, use of thirdgeneration cephalosporins active against Pseudomonas aeruginosa, and female sex.
Earlier studies also found the use of third-generation cephalosporins active against P. aeruginosa to be a risk factor for SM-VAP (26,27). Thus, in the study by Hanes et al., the administration of cefepime was signi cantly associated with the occurrence of SM-VAP (OR 3.31 [1.12 -9.72]). Interestingly, the number of patients evaluated in that study was small, which reinforces the association between the two events (26). More recently, Imoto et al. found a signi cant association between the use of third-generation cephalosporins active against P. aeruginosa and the occurrence of hospital-acquired pneumonia caused by SM (27).
Our analysis found no association between the use of carbapenems and the occurrence of SM-VAP. By contrast, in their 2019 case-control study of 102 patients, Ibn Saied et al. identi ed exposure to carbapenems as a risk factor for this infection (OR 3.20 [1.77-5.79]). We initially explained the difference between these and our ndings by the fact that the use of carbapenem was lower in our patients since only rst episodes of VAP were included in our analysis. Yet, on closer examination, we found that the percentage of patients who received carbapenems in our study was comparable to that in the study by Ibn Saied et al. (8).
No association was found in our study between the use of aminoglycosides and the occurrence of SM-VAP. While Van Couvenberghe et al. did nd an association between these two events in their 1997 study, their sample was small and included only 25 cases of SM-VAP (28).
In line with published data, we found no association between other types of antibiotic therapy and the occurrence of SM-VAP. However, some studies identi ed exposure to uoroquinolones as a risk factor for hospital-acquired pneumonia caused by NF-GNB (29,30), and others highlighted an association between this antibiotic and multidrug-resistant SM infection (31)(32)(33). While the use of glycopeptides has been identi ed as a risk factor for bacteremia caused by NF-GNB (34), to our knowledge no association has been reported between this antibiotic and the occurrence of SM-PAVM. In our study, few patients received glycopeptides due to the absence of methicillin-resistant S. aureus and, more generally, due to the rarity of this microorganism in the ecology of our ICU. Interestingly, no association was found between glycopeptides and SM-VAP in the study by Ibn Saied et al., whose population was similar to ours yet was more frequently treated with this antibiotic (8).
To date, no association has been reported between malnourishment and SM-VAP. However, some studies suggest that parenteral nutrition is a risk factor for SM infection (8,35). Others indicate that the use of parenteral nutrition in ICU has been declining in recent years and currently concerns less than 10% of ICU patients (36). In our study, 9% of patients required parenteral nutrition during their stay in ICU. While we found parenteral nutrition to be signi cantly higher in patients with SM-PAV, we were unable to determine the impact of this treatment on the occurrence of the infection.
Rectal colonization with SM was not a risk factor for SM-VAP in our patients, which is in line with published data. However, one study found the risk of developing an infection with SM to increase in cases of relative abundance in the oral area (37) -a parameter that was not analyzed in our study. More generally, several studies have highlighted an association between digestive colonization with the NF-GNB Acinetobacter baumannii or Pseudomonas aeruginosa and infection in ICU (38)(39)(40).
In our study, chronic renal failure and chronic respiratory failure were independent risk factors for SM-VAP. No such association has been reported in the literature to date. Our nding may be explained by the high prevalence of chronic renal failure in our study population (16.9%). In a recent comparable study, this prevalence varied between 4.5 and 7.1% depending on the group (8). Note that the prevalence of diabetes in our study population was also high at 34.3%, which is much higher than the prevalence reported for the general French population (41).
No association has been reported to date between chronic respiratory disease and the occurrence of SM-VAP. It is established, however, that patients with chronic respiratory disease such as cystic brosis and COPD are more likely than the general population to carry multidrug-resistant microorganisms, and in particular NF-GNB. In cystic brosis patients, the most common microorganism is P. aeruginosa and other NF-GNB are increasingly being detected (42)(43)(44)(45). Fewer data are available on the bacterial ecology of patients with COPD. We found only one study comparing patients with and without COPD who developed VAP during their stay in ICU: while the incidence of VAP was the same in both groups, that of E. coli and SM was signi cantly higher in patients with COPD (46). In our study, the incidence of a rst episode of SM-VAP was 0.93%. This is higher than gures reported in the literature, with two recent studies reporting incidences of 0.27 and 0.48%, respectively (8,9). While the incidence of SM in the respiratory tract was 2% in the study by Nseir et al., this gure included both colonization and infection cases (7). Some studies have shown that patients living in tropical environments are at greater risk of developing pneumonia caused by NF-GNB, and in particular by A. baumannii (47,48). Interestingly, pneumonia caused by A. baumannii was common in our control group, as it represented 5.8% of monomicrobial infections and 11.9% of polymicrobial infections. One study conducted in Reunion Island reported a high prevalence of A. baumannii in non-human reservoirs outside hospital and suggested a link with the high prevalence of pneumonia caused by A. baumannii in the Reunionese population (49). This could explain not only the high incidence of infections with A. baumannii in our study population, but also that of infections with SM given the similarities between the two microorganisms.
In our study, almost half of patients with a rst episode of SM-VAP (49.4%) died in ICU, and an even higher percentage (57.3%) died before hospital discharge. This nding is in line with published data, with studies reporting an ICU mortality ranging from 41.3 to 50%. (9,12,13,50,51). However, in the 2002 study by Hanes et al., ICU mortality (23.1%) was much lower than in our study (26). This difference may be explained by the fact that their study population (median age of 47 years) was much younger than ours (median age of 60 years) and was composed of severe trauma patients with fewer comorbidities. Other studies also reported lower ICU mortality, but for smaller cohorts (6,7).
The excess mortality due to SM in our patients may be explained by delayed initiation of effective antibiotic therapy. Indeed, probabilistic antibiotic therapy was active against SM in only 66.3% of our patients with SM-VAP, whereas it was active against the identi ed microorganism in 94.2% of patients in the control group. Several authors have highlighted an association between ineffective antibiotic therapy and mortality in patients with hospital-acquired pneumonia caused by SM (12,16,17,19). In particular, some studies on SM-VAP have reported signi cantly higher mortality in patients receiving initial antibiotic therapy inactive against SM (6,7,25).
In patients with polymicrobial SM-VAP, probabilistic antibiotic therapy was effective against the coinfecting microorganism in 100% of cases and against SM in less than 50% of cases (18/39). Interestingly, studies on infection with SM report excess mortality in patients co-infected with P. aeruginosa (52). In our study, P. aeruginosa was the most common co-infecting microorganism (28.2% of patients with polymicrobial SM-VAP), which could also explain, at least in part, the high mortality observed in our study population.
The main limitation of our study is its retrospective nature, which could have introduced biases. Since our study was conducted in a single center, our results cannot necessarily be extrapolated to other centers.
While our diagnostic criteria were very strict, many of the evaluated cases of SM-VAP were polymicrobial infections, which may have led to a bias in the analysis of risk factors for developing a rst episode SM-VAP. Another limitation of our study is the high percentage of patients with co-morbidities (especially cardiovascular) compared to other studies on SM-VAP.

Conclusion
In our study, chronic respiratory disease, chronic renal failure, use of third-generation cephalosporins active against P. aeruginosa, and female sex were identi ed as risk factors for a rst episode of SM-VAP. Moreover, SM was associated with excess mortality, as was the case in other studies on hospital-acquired pneumonia. This excess mortality could be explained by delayed initiation of effective antibiotic therapy, itself due to the resistance pro le of SM and the di culty in identifying patients at risk of infection.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
No source of funding to declare for this research.
Authors' contributions CC collected, analyzed, interpreted the data from this research and was a major contributor in writing the manuscript. BP collected, analyzed, interpreted the data from this research and was a major contributor in writing the manuscript. LT collected the data from this research. NA analyzed and interpreted the data from this research.