Morbidity and mortality related to pneumonia and TRACHEOBRONCHITIS in ICU after lung transplantation

Background Bacterial respiratory infections (BRI) are major complications contributing to increased morbidity and mortality after lung transplantation (LT). This study analyzed epidemiology and outcome of 175 consecutive patients developing BRI in ICU after LT between 2006 and 2012. Methods Three situations were described: colonization determined in donors and recipients, pneumonia and tracheobronchitis during the first 28 postoperative days. Severity score, demographic, bacteriologic and outcome data were collected. Results 26% of donors and 31% of recipients were colonized. 92% of recipients developed BRI, including at least one episode of pneumonia in 19% of recipients. Only 21% of recipients developed BRI with an organism cultured from the donor’s samples, while 40% of recipients developed BRI with their own bacteria cultured before LT. Purulent sputum appears to be an important factor to discriminate tracheobronchitis from pneumonia. When compared to patients with tracheobronchitis, those with pneumonia had longer durations of mechanical ventilation (13 [3–27] vs 3 [29], p = 0.0005) and ICU stay (24 [16–34] vs 14 [9-22], p = 0.002). Pneumonia was associated with higher 28-day (11 (32%) vs 9 (7%), p = 0.0004) and one-year mortality rates (21 (61%) vs 24 (19%), p ≤ 0.0001). Conclusions These data confirm the high frequency of BRI right from the early postoperative period and the poor prognosis of pneumonia after LT.

Although recent guidelines have defined pneumonia, tracheobronchitis and colonization after lung and heart transplantation [8], the distinctions between these entities are blurred by several confounding factors such as ischemia-reperfusion syndrome, immunosuppression or atelectasis [9]. Although a consensus has been reached concerning the treatment of bacterial pneumonia [8], the need to extend antibiotic therapy to tracheobronchitis remains controversial. Criteria usually proposed for the diagnosis of postoperative pneumonia in LT recipients are difficult to identify, resulting in a high risk of missing pneumonia in these immunocompromised patients [8]. Our local policy is to initiate antibiotic therapy as soon as a patient is suspected of having BRI, even in the absence of criteria of pneumonia.
The aim of the present study was to compare the characteristics of patients who developed one or more episodes of BRI during the first 28 days after LT in ICU with those who did not. The influence of colonization of the bronchial tree and tracheobronchitis on the development of pneumonia and the role of these entities on the outcome were also evaluated.

Study population
Between January 2006 and December 2012, all consecutive adult patients undergoing LT and admitted to the intensive care unit (ICU) for postoperative care for at least 24 h were retrospectively included in a database. The study period included the length of ICU stay during the first 28 days following transplantation. According to French law, no informed consent was required in view of the observational and retrospective nature of this study. The study was approved by the Paris North Hospitals Institutional Review Board (Paris VII University, AP-HP, IRB No. 00006477).

Perioperative management
Surgical transplantation procedure is standardized according to our local policy [10,11]. Perioperative care, including postoperative management, was standardized for all patients according to our standard protocol [12]. Immunosuppressive therapy was based on a combination of prednisolone, ciclosporin and azathioprine subsequently replaced by mycophenolate mofetil.
In patients without previous colonization of the respiratory tract, perioperative antibacterial prophylaxis consisted of cefamandole at induction of anesthesia and continued for 48 h. In patients with preoperative colonization, prophylaxis was adapted to the microorganisms isolated. Our center's policy does not include digestive or bronchial decontamination prior to LT.

Microbiologic samples
Respiratory colonization samples were performed in a large proportion of patients more than 6 months before transplantation. Respiratory samples are not routinely performed immediately prior to transplantation in our center.
Pretransplant tracheal bacterial colonization was recorded. Donor airway samples were obtained by bronchial aspiration during the surgical procedure before transplantation.
During the postoperative period, fiberoptic bronchoscopy was performed daily with invasive respiratory tract samples collected by bronchial aspiration (BA) every two or three days and by bronchoalveolar lavage [13] in the case of suspected viral or fungal infection. Microbiologic samples were processed according to the usual techniques. Susceptibility testing was performed as previously described [14].

Definitions of BRI
Pneumonia was defined as positive culture of BA samples yielding ≥10 5 colony-forming units (CFU)/mL combined with new or persistent lung infiltrate on chest X-ray, and two or more of the following criteria: temperature ≥ 38.4°C or < 36°C, WBC count >11,000 /mm 3 or <4000 /mm 3 , at least 30% decrease of PaO 2 / FiO 2 ratio, and purulent secretions [15]. Tracheobronchitis was defined as positive culture of microbiologic samples, normal appearance or moderate interstitial infiltrates on chest X-ray and at least one of the previously described clinical signs [16]. Colonization was defined as positive culture of microbiologic samples with no clinical, laboratory or radiological signs [8].

Antibiotic therapy
According to local policy, all cases of suspected postoperative BRI (pneumonia or tracheobronchitis) were treated by empiric antimicrobial therapy, followed, when the diagnosis was confirmed, by targeted therapy according to susceptibility testing for a total duration of seven days. Our local empiric antimicrobial therapy consisted of a combination of cefepime and amikacin.

Data collection
Demographic characteristics, underlying diseases and clinical data were recorded. Severity scores (SOFA and SAPS II) were assessed on ICU admission. At the time of pulmonary samples, clinical characteristics (purulent secretions, abundance of secretions, temperature), laboratory parameters (white blood cell (WBC) count, PaO2), chest X-ray imaging and results of microbiologic cultures were collected. Morbidity criteria (ICU length of stay, duration of mechanical ventilation, primary graft dysfunction score) and 28-day and one-year mortality after LT were assessed [17].

Statistical analysis
All statistical analyses were performed using GraphPad Prism 5.0 (Apple, Cupertino, CA). Quantitative variables are expressed as median and interquartiles (25th-75th percentiles) and were compared by a nonparametric Mann-Whitney test. Categorical variables are expressed as proportions and absolute numbers, and were compared by Chi-square test or Fisher's exact test. In the subgroup of patients with BRI, we compared the clinical presentation, microbiologic data and outcome of patients with and without criteria of pneumonia. Risk factors for one-year mortality were analyzed by univariate analysis. Variables with a p-value< 0.15 on univariate analysis were entered in a multivariate logistic regression model. One-year mortality was estimated by the Kaplan-Meier method. A p-value < 0.05 was considered statistically significant.

Demographic and clinical characteristics
Between 2006 and 2012, 175 LT were performed. Demographic data are presented in Table 1.
During the first 28 days after LT, 161 (92%) patients developed at least one episode of BRI. A median of 2 [1][2][3] episodes of BRI were observed per patient for a total of 380 episodes. In this population, 127 (73%) patients developed tracheobronchitis with no episode of pneumonia and 34 (19%) patients developed one or more episodes of pneumonia. Fourteen (8%) patients did not experience any BRI.

Bacterial colonization
Forty-seven donors and 55 recipients were colonized before transplantation. Bacterial species involved in colonization are presented in Table 3.
No significant difference in the frequency of pneumonia due to organisms cultured from the host or the donor was observed between single and double lung transplantation (4/20 patients versus 5/14 patients, respectively, p = 0.43).

Clinical characteristics of the patients in the various subgroups
Apart from age, no significant differences were observed between patients with and without BRI in terms of demographic or severity characteristics (Table 1).
No significant demographic differences were observed patients with pneumonia and patients with tracheobronchitis with no episode of pneumonia apart from the need for ECLS (Table 1), while they presented different clinical features at the time of pulmonary sampling (Table 4).

Outcome
Patients with at least one episode of pneumonia had higher 28-day and one-year mortality rates (Fig. 3) and longer duration of mechanical ventilation and ICU stay than patients with tracheobronchitis without pneumonia (Table 5).

Discussion
This study, based on a large cohort of 175 LT patients, determined the epidemiology, prognosis and outcome after BRI in the early postoperative period following LT. Donor and recipient colonization played a limited role in the development of BRI. BRI resulted in pneumonia  Fig. 1 Relationship between pneumonia and recipient-donor colonization due to the same micro-organism in 17% of cases. This study also confirms the poor prognosis of pneumonia after LT.

Bacterial species/colonization
Enterobacteriaceae, Streptococcus spp. and Pseudomonas aeruginosa were the predominant species isolated in our cohort, in line with other studies analyzing the postoperative period after LT [1,5,18]. Only 21% of recipients developed BRI with a microorganism cultured from donor samples. The majority of these pathogens were Gram-positive bacteria (especially staphylococci and streptococci) and Enterobacteriaceae.

Definition of pneumonia and tracheobronchitis
All patients with a diagnosis of BRI in our center received antibiotics adapted to the species cultured. Our antibiotic policy differs from that described in several other published studies, in which only patients with pneumonia were treated by antibiotics [4], resulting in a lower selection pressure, but an increased risk of undertreating true pneumonia, and increased mortality rates, as reported by Iregui et al. [18]. Percentage adequacy in our population, published in a previous paper, demonstrated a high rate of MDR bacteria in cases of late pneumonia [14].
It can be very difficult to distinguish between tracheobronchitis and pneumonia during the postoperative period following LT. In 2010, the International Society of Heart and Lung Transplantation Infectious Diseases Council working group defined proven pneumonia in cardiothoracic transplant recipients on the basis of clinical, radiologic and laboratory criteria. According to the authors, this definition applied not only to the early phase of LT, but also to the subsequent phase. However, during this crucial LT postoperative period, several criteria for the diagnosis of pneumonia may be difficult to assess.
We observed the presence of these criteria in both tracheobronchitis and pneumonia. Low rates of fever or hypothermia can be easily explained by administration of steroids and immunosuppressive agents. High rates of leukocytosis/leukopenia are frequently reported in the LT postoperative period and are not specific to sepsis.  For example, ischemia-reperfusion is an important confounding factor, especially during the first five postoperative days [19,20]. Similarly, the characteristics, changes and quality of sputum are difficult to assess during the LT postoperative period. Ischemia-reperfusion syndrome and fluid administration during and after surgery can make sputum characteristics difficult to interpret. However, purulent sputum appeared to be a relevant factor in our cohort of patients. Cough, dyspnea, tachypnea, or pleural rub, rales or bronchial breath sounds are also nonspecific signs and appear to be totally inappropriate and of limited value during the postoperative period, especially in ventilated patients. While the PaO 2 /FiO 2 ratio appears to be a relevant parameter in ICU patients for the diagnosis of VAP, this criterion can be misleading after LT due to the presence of ischemia/reperfusion syndrome, primary graft dysfunction, and atelectasis or lung collapse. Finally, chest X-ray changes are challenging issues in these patients with ischemia/reperfusion, pulmonary edema, atelectasis and lung collapse, pleural effusion or acute rejection. X-ray signs of infiltration could be a very sensitive criterion for pneumonia in these patients, but are associated with poor specificity, as illustrated by the high rate of these signs in patients with tracheobronchitis.

Links between BRI, tracheobronchitis, pneumonia and outcome
Riera et al. recently described an increased mortality rate in patients with postoperative pneumonia after LT [1].
ICU mortality increased to 40% and the in-hospital mortality rate was 50% in patients who developed postoperative pneumonia. This finding highlights the severity of postoperative pneumonia, even when appropriate antibiotics are rapidly prescribed. Although some studies have not found any link between pneumonia and mortality [21,22], mortality rates remain high in this disease, which can be explained by the severe immunocompromised status of these patients, especially during this early phase of the immunosuppressive protocol, with potentially limited access of antibiotics to the lung during this crucial period.
In this context, we compiled all cases of pneumonia in our cohort according to the definition proposed by Riera et al. and identified 42 episodes of pneumonia in 34 patients. As described by Riera et al., the major finding is that, compared to tracheobronchitis, postoperative pneumonia is associated with dramatically increased 28day (32%) and one-year (61%) mortality.
With a 28-day mortality of 7% and a one-year mortality of 19%, the outcome of patients with tracheobronchitis with no episode of pneumonia did not differ from that reported in other studies conducted during the LT postoperative period [1,18]. However, no significant difference in terms of demographic and microbiologic data was observed between patients with one or more episodes of pneumonia and patients with tracheobronchitis but no episode of pneumonia. In view of the high mortality rate in this context, it is essential to rapidly distinguish between these two entities in order to focus on patients with pneumonia. Purulent sputum seems to be a very strong and useful clinical criterion to distinguish tracheobronchitis from pneumonia. However, it is a very subjective decision to note sputum as purulent. Further studies are required to characterize this group of patients with poor outcome.
Lastly, only a small proportion of patients (n = 14) did not develop any form of lung infection during the early postoperative period. No general characteristics were identified to differentiate these patients from those with BRI. However, the mortality rate of this small subset of patients (21% at 28 day and 42% at one year) remains high. This high 28-day mortality rate can be explained  by non-infectious causes, such as primary graft dysfunction or surgical complications, while infectious diseases, especially late pneumonia, were mainly responsible for the high one-year mortality rate.

Therapeutic consequences of our findings
The findings of this study cannot be used to determine whether our local policy to treat all episodes of tracheobronchitis was able to decrease the incidence of earlyonset pneumonia in our population. Most episodes of pneumonia were due to species unrelated to the donor's or recipient's colonization flora, as also observed in other studies in lung transplant cohorts. For example, in the study by Riera et al. analyzing 170 patients, most cases of pneumonia were totally independent of colonization prior to the graft [1]. Riera et al. observed only 3/20 cases of pneumonia caused by species cultured prior to the graft. In Riera's study, gastroparesis was significantly associated with both pneumonia and ventilator-associated tracheobronchitis. Although most emerging organisms were not cultured from the host's respiratory tract, we cannot exclude a host digestive tract origin of these bacteria. The gastric reservoir is a wellknown source of contamination, which has led many teams to propose the use of selective decontamination.

Limitations
Our study presents a number of limitations. Firstly, this single-center study reflects our local practice and essentially concerned COPD and idiopathic pulmonary fibrosis (IPF) patients. The epidemiologic findings would therefore probably be different in cystic fibrosis patients. Secondly, bacterial resistance was not evaluated in this study, but could be a relevant parameter, especially as all cases of BRI in our cohort were treated. We have previously reported a progressive increase of resistance rates, which could be related to our local antibiotic policy [14].
Thirdly, analysis of the inflammatory status was not included in our decision-making process, which is based exclusively on microbiologic data. The specific host inflammatory response could be an important trigger to stratify the development of tracheobronchitis or pneumonia. However, we have previously shown that the early inflammatory response is a confounding factor for the diagnosis of early pneumonia [12].
Fourthly, the pathogenic role of enterococci and coagulase-negative staphylococci (coNS) in pneumonia remains controversial, at least in immunocompetent ICU patients [23]. Enterococci and coNS BRI were treated according to our local rules. However, this policy is subject to debate and could be considered to be excessive.
Fifthly, respiratory colonization samples performed in a large proportion of patients more than 6 months before transplantation may have been a source of bias. Information about species during the last 6 months in our practice might be interesting, especially to determine the most appropriate antibiotic prophylaxis during the first 48 h following transplantation.
Lastly, this was a retrospective study and prospective evaluation of our practice could be a useful approach in future studies.