The contribution of Pseudomonas aeruginosa infection to clinical outcomes in bronchiectasis: a prospective cohort study

Abstract Objectives The impact of Pseudomonas aeruginosa on the prognosis of bronchiectasis remains controversial. This study aimed to explore the prognostic value of P. aeruginosa in adult patients with bronchiectasis in central-southern China. Patients and methods This prospective cohort study enrolled 1,234 patients with bronchiectasis between 2013 and 2019. The independent impact of P. aeruginosa on all-cause mortality, annual exacerbations, and hospitalizations was assessed. Results P. aeruginosa was isolated from 244 patients (19.8%). A total of 188 patients died over a follow-up period of 16 (1–36) months. Patients with P. aeruginosa had a longer disease course, poorer lung function, more lung lobe involvement, and more severe Bronchiectasis Severity Index (BSI) stage than those without P. aeruginosa. The independent impact of P. aeruginosa was observed on frequent hospitalizations but not on mortality and frequent exacerbations. Moderate- or high-risk comorbidities increased the risk of mortality (hazard ratio [HR]: 1.93, 95% confidence interval [CI]: 1.26–2.95), and this effect was magnified by the presence of P. aeruginosa (HR: 2.11, 95% CI: 1.28–3.48). Conclusions P. aeruginosa infection acts as a marker of disease severity as well as predictor of frequent hospitalizations. P. aeruginosa had no independent effect on all-cause mortality. P. aeruginosa combined with moderate- or high-risk comorbidities posed an increased risk of mortality. The management of comorbidities may be a critical target during the treatment of P. aeruginosa infection in bronchiectasis. KEY MESSAGE: P. aeruginosa increased the risk of frequent hospitalizations; however, it had no independent impact on all-cause mortality. P. aeruginosa combined with moderate- or high-risk comorbidities posed an increased risk of mortality. The management of comorbidities may be a critical target during the treatment of P. aeruginosa infection in bronchiectasis.


Introduction
Non-cystic fibrosis bronchiectasis (hereafter referred to as bronchiectasis) is a chronic progressive bronchus or bronchiole dilation due to complex interactions between recurrent infection, unbalanced immune regulation, impaired mucociliary clearance, and progressive airway structure damage or obstruction [1]. The estimated prevalence of bronchiectasis is 52.3 cases per 100,000 in the United States [2] and 67 cases per 100,000 in Germany [3], and the prevalence increases each year by 8.7% [4]. The crude annual mortality rate of bronchiectasis varies between 2% and 10% [5][6][7], which continues to pose a heavy disease burden on both developing and developed countries.
Pseudomonas aeruginosa (P. aeruginosa) tops the list of organisms isolated from sputum in patients with bronchiectasis, either during stable state or exacerbation [8,9]. It is usually considered to be highly responsible for the poor clinical outcomes of bronchiectasis. As bronchiectasis is a heterogeneous disease, the aetiology, microbial spectrum, and disease severity vary under different geographic and economic background [10]. Some studies have suggested that chronic infection with P. aeruginosa independently affects clinical outcomes, including hospitalizations, exacerbations [11,12] and mortality [12,13], in bronchiectasis, whereas other studies have drawn contrasting conclusions [5,14,15]. It remains controversial whether P. aeruginosa has an independent effect on clinical outcomes or is simply a marker of disease severity.
In China, bronchiectasis has long been neglected, and studies regarding clinical profiles and the impact of P. aeruginosa infection on its prognosis are scanty, mainly concentrated in two developed metropolises [12,16]. Therefore, we aimed to explore the clinical profiles of P. aeruginosa infection in bronchiectasis exacerbation and its effect on all-cause mortality, future hospitalizations, and exacerbations in a large tertiary hospital in central-southern China.

Study design and participants
This prospective cohort study collected data on patients with bronchiectasis exacerbation between 2013 and 2019 at the Second Xiangya Hospital of Central South University. A flowchart of patient enrolment is displayed in Figure 1. Consecutive adult patients with a diagnosis of bronchiectasis based on high-resolution computed tomography (HRCT) scans and presenting corresponding respiratory symptoms attributable to bronchiectasis exacerbation were enrolled according to the guidelines [17]. Patients were excluded if they lacked HRCT scan results or records of sputum culture in the hospital. Only the first medical records of patients with multiple admissions were reviewed for our analysis. This study was approved by the institutional review board of the Second Xiangya Hospital in China (luoh202006), and informed consent was obtained from all participants.

Data collection
Data on demographics (age, sex, occupation, height, and weight), disease course, symptoms, aetiology, comorbidities, forced expiratory volume in the first second (FEV1% predicted), radiologic findings, longterm treatment of macrolides, and Medical Research Council (MRC) dyspnoea score were collected. The aetiology and diseases associated with bronchiectasis were identified according to Spanish guidelines [18].
The Bronchiectasis Aetiology Comorbidity Index (BACI) [19], which is a sum score (range, 0-55) of 13 weighted diseases, was used to assess comorbidities, where higher scores denote an increasing burden of comorbidities specified in bronchiectasis. Patients were classified into the following tertiles: low-risk comorbidities (for patients with a score of 0), moderate-risk comorbidities (for patients with scores !1 and <6), and high-risk comorbidities (for patients with a score !6).

Sputum microbiology and group assignment
All microbiological tests were performed on spontaneous sputum samples. Sputum samples were considered eligible if they contained more than 25 leukocytes and less than 10 squamous cells per lowpowered field (10 Â 10). Specific microorganism infection was considered if sputum cultures were positive on one or more occasions and accorded with clinical physician judgement.
We assigned patients into the following two groups: (1) the PA group for those with P. aeruginosa isolation from sputum and (2) the non-PA group for those without P. aeruginosa isolation from sputum.

Follow-up and clinical outcomes
Patients were followed up after discharge via specifically designed telephone interviews every 1-3 months. Follow-up was completed on 30 June 2020. There was a median follow-up duration of 16 (1-36) months, in which the overall number of exacerbations and hospitalizations due to exacerbations during the follow-up period were recorded, and mortality was assessed.
The clinical outcomes in this study were all-cause mortality, frequency of annual exacerbations, and frequency of annual hospitalizations due to exacerbations.
An exacerbation of bronchiectasis was defined as at least three of the following symptoms: cough frequency, sputum volume and/or consistency, sputum purulence, dyspnoea and/or exercise tolerance, fatigue and/or malaise, and haemoptysis; where the condition deteriorated for at least 48 h, and beyond the daily variation range, a change in treatment was required [21]. Patients were considered to have frequent exacerbations if there were two or more annual exacerbations (total number of exacerbations divided by follow-up years) and frequent hospitalizations if there were one or more annual hospitalizations (total number of hospitalizations divided by follow-up years).

Statistical analysis
Descriptive statistics of demographic and clinical variables are presented as median (interquartile range) or counts (n) and proportions (%), unless otherwise stated. Data comparisons were applied between the PA and non-PA groups. The chi-squared test was used for categorical data. The student's t-test was used for continuous variables showing normal distributions and the Kruskal-Wallis test for those with non-normal distributions. Unless stated otherwise, the significance level was set at p < .05.
Kaplan-Meier curves were constructed to illustrate survival data, and Cox proportional hazard regression analysis was used for univariate and multivariable mortality analysis to estimate hazard ratios (HRs) and their 95% confidence intervals (CIs). Similarly, univariate and multivariate logistic regression analyses were performed to identify variables associated with bronchiectasis hospitalizations and exacerbations, and odds ratios (ORs) and their 95% CIs were calculated. We included the following baseline variables when studying potential in the univariate analyses: sex; age; BMI; smoking habit; symptoms of haemoptysis; exacerbations in the previous year; MRC dyspnoea score; Reiff score; FEV1% predicted; BACI score; aetiology and associated diseases (unknown aetiology, chronic respiratory disease; connected tissue disease, postinfectious); long-term oral macrolides; sputum microbiologic culture (P. aeruginosa, Acinetobacter baumannii, Klebsiella pneumoniae, Escherichia coli, Haemophilus influenzae, Staphylococcus aureus, Candida, and Aspergillus fumigatus); BSI stage; status of PA; and comorbidities (only for survival analysis). Each variable was initially tested individually before we added all variables that showed a univariate association (p < .10) to the multivariable model, except for sex and age, which had to appear in both models. Backward stepwise selection (likelihood ratio) (p in < 0.05 and p out > 0.10) was used to determine the factors associated with mortality, hospitalizations, and exacerbations. The Hosmer-Lemeshow goodness-of-fit test was performed to assess the overall fit of the final model. All data were analysed using SPSS version 26.0 (SPSS Inc., Chicago, IL, USA).
We hypothesised that the association of P. aeruginosa with mortality may be comorbidity-dependent. Based on the condition of the P. aeruginosa infection andBACI score, we reclassified the patients into four subgroups as follows: (1) non-PA with low-risk comorbidities (n ¼ 350); (2) non-PA with moderate-or highrisk comorbidities (n ¼ 640); (3) PA with low-risk comorbidities (n ¼ 78); and (4) PA with moderate-or high-risk comorbidities (n ¼ 166). The survival curves       of the four subgroups are shown in Figure 3. In the third multivariable analysis model of subgroups (

Discussion
To the best of our knowledge, this is the largest Chinese single-center study on clinical profiles and outcomes of P. aeruginosa infection in bronchiectasis.  Our study demonstrates that P. aeruginosa increased the risk of frequent hospitalizations in bronchiectasis but had no independent role in all-cause mortality and frequent exacerbations. Moderate-or high-risk comorbiditiesincreased the risk of all-cause mortality, and the risk could have been magnified by P. aeruginosa. P. aeruginosa was the most common organism (19.8%) isolated from the sputum of bronchiectasis patients, which is consistent with the findings of some previous reports [5,16,22] yet different from thoseof others [10,13,14]. The reason for these disparities may be the difference in the definition of P. aeruginosa infection, geographic and economic background, and race.
In our study, patients with P. aeruginosa had many different clinical profiles from those without P. aeruginosa, such as a longer clinical course, poorer lung function, higher rate of pulmonary hypertension, more radiologic involvement, and more frequent future hospitalizations, indicating that P. aeruginosa is a marker of disease severity.
The all-cause mortality was 15.2% with a mean follow-up of 16 months in our cohort, which is higher than that reported by Araujo et al. (10.8% in 5 years) and Wang et al. (7.6% in 44 months) [12,14]. The reason for the higher mortality rate in this cohort might have been the complex comorbidities in these patients, considering that 40.1% of the patients had moderate-risk comorbidities and 25.2% high-risk comorbidities. Another reason might have been the severe disease stage of the patients, as the majority of these patients fell into the moderate or severe BSI stage.
The effect of P. aeruginosa on the risk of mortality in patients with bronchiectasis is controversial. Several studies have reported that chronic infection with P. aeruginosa increases the risk of mortality in bronchiectasis patients [5,13,23]; however, Chalmers et al. [11] and Araujo et al. [14] analysed data from a large cohort of 2,596 patients and found that P. aeruginosa had no independent impact on mortality. Previous studies have assessed the impact of certain comorbidities, such as COPD, asthma, and connective tissue disease, without a systematic review of comorbidities [5,6,11,12,14,15]. An international, multicenter cohort analysis of outpatients with bronchiectasis from Europe utilized the BACI score to assess comorbidities comprehensively and demonstrated that the BACI score could predict mortality; nevertheless, this study did not explore the compound effect of comorbidities with other clinically important factors, such as P. aeruginosa [19]. In our cohort, we found that the independent predictors of all-cause mortality were age, low BMI and BACI score, and BSI stage but not P. aeruginosa infection. In the subgroup analysis, we found that moderate-or high-risk comorbidities increased the risk of all-cause mortality. Furthermore, P. aeruginosa magnified the risk of mortality in patients with moderate-or high-risk comorbidities. These data suggest the importance of managing comorbidities in all bronchiectasis patients, especially in cases of moderate-or high-risk comorbidities combined with P. aeruginosa presence.
Our study showed that P. aeruginosa infection was independently associated with frequent hospitalizations due to exacerbations, which is consistent with the findings of a previous study [11,12,14,15,24]. On the other hand, our study found no independent association of P. aeruginosa infection with frequent exacerbations, contradicting the findings from the previous study [11,12,14,15]. As oral antibiotics and mucolytic are available in many places in China [25], bronchiectasis patients often take self-prescribed antibiotics irregularly to suppress the symptoms instead of consulting doctors, which potentially underestimates the exacerbation rates and may be the reason we failedto observe the independent effect of P. aeruginosa on frequent exacerbations. Hospitalizations signify a more severe disease condition, heavier disease burden, and high expenditure, despite different admission criteria in different areas. The eradication of P. aeruginosa colonization may reduce hospitalization and disease burden in the bronchiectasis population. Recently, the eradication of P. aeruginosa has focussed on antibiotic therapy [18,26]. Limited randomised controlled studies have shown that long-term oral macrolides can prevent bronchiectasis exacerbations [27][28][29]. As bronchiectasis progression is a complex interaction of the host, environment, and microorganisms, interventions for P. aeruginosa infection in bronchiectasis are not limited to antibiotics. Host immune regulation, including passive and active immunity, and pulmonary rehabilitation, especially the airway clearance technique, have been utilized in individual treatments [30]. Large randomized controlled trials are still required to clarify the management of bronchiectasis with P. aeruginosa infection.
Our study has several limitations. First, this was a prospective study in a single-center tertiary hospital, where we could not define the PA group as chronic P. aeruginosa colonization because we were unable to perform subsequent repeated sputum tests to confirm their state of bacterial colonization. It would be valuable to trace multiple cultures in future longitudinal studies to assess chronic P. aeruginosa colonization, and large prospective cohort or randomized controlled trials are warranted to further evaluate the influence of P. aeruginosa infection on mortality and exacerbations. Second, we described microbiologic features based on sputum culture on account of its availability, which may underestimate microbial diversity, especially in viruses. Metagenomics sequencing of respiratory specimens may provide more insights into microbes in the future [31]. Third, we did not analyse the impact of the P. aeruginosa resistance pattern and phenotype/strain type on clinical outcomes. The existence of multidrug-resistant P. aeruginosa (MDR-PA) isolates is an emergent health threat worldwide. However, evidence on the association between MDR-PA isolates and bronchiectasis prognosis is limited [16] and MDR-PA needs to be investigated further in future studies. Whereas a cross-sectional study of mucoid P. aeruginosa and non-mucoid P. aeruginosa in bronchiectasis presented different hospitalizations and exacerbations in the previous year [32], analyses of the prognostic effects of mucoid P. aeruginosa and nonmucoid P. aeruginosa on bronchiectasis are limited. Therefore, more well-designed prospective studies are required to explore the relationship between P. aeruginosa resistance patterns and phenotype/strain type on clinical outcomes. Additionally, we did not consider infections by nontuberculous mycobacterium (NTM) in this study because most patients did not undergo examinations for NTM. NTM infection in bronchiectasis is essential for full assessment, and it is critically important to perform NTM cultures in bronchiectasis in future clinical management.
In conclusion, our study provides a more comprehensive clinical view of the role of P. aeruginosa infection in disease severity and burden. P. aeruginosa infection in bronchiectasis indicated longer disease courses, poorer lung function, and more severe radiologic involvement. P. aeruginosa increased the risk of frequent hospitalizations but had no independent role in all-cause mortality and frequent exacerbations in bronchiectasis. Moderate-or high-risk comorbidities increased the risk of all-cause mortality, and the risk was potentially magnified by the presence of P. aeruginosa. Therefore, P. aeruginosa infection acts as a marker of disease severity as well as predictor of frequent hospitalizations. The management of comorbidities in bronchiectasis may be a critical target during the treatment of P. aeruginosa infections.