Objective: To identify known and unknown bacterial pathogens in patients with pleural infections using a combination of conventional culture and next-generation sequencing approaches.

Hypothesis to be tested: The investigators hypothesize that next-generation sequencing will serve as a comprehensive approach to identify culturable and unculturable bacterial pathogens in patients with pleural infections compared to the conventional culture.

Design and subjects: This is a prospective cohort study to be conducted in the medical department of a tertiary care hospital in Hong Kong involving patients with pleural infection. Patients will be recruited if a pleural infection is suspected, with pleural fluid sampled and a 6-month follow-up. The clinical management by the medical team will not be interfered.

Study instruments: Pleural fluid will be collected for conventional culture, 16S amplicon and shotgun metagenomic sequencing in parallel. The clinical information will be collected to clarify the causative correlation between symptoms, clinical outcomes and pathogen infections.

Main outcome measures: The full spectrum of causative bacteria in pleural infection will be characterized. The diagnostic performance of identifying causative bacteria in pleural infection will be compared between the studied methods. The antimicrobial resistance pattern, clinical outcomes of pleural infection will also be compared between groups as categorized by the pattern of bacteriology identified by different methods.

Data analysis: With reference to the conventional culture as the gold standard, sensitivity, specificity, positive predictive values and negative predictive values of 16S ribosomal ribonucleic acid (rRNA) gene amplicon and shotgun metagenomic sequencing will be calculated.

Parapneumonic effusion (PPE) is a common and serious clinical problem. Among patients suffering from community-acquired pneumonia, up to 57% of them would develop parapneumonic effusion (PPE), and 7.2% would develop complicated parapneumonic effusion (CPPE) and empyema. The latter two, collectively known as pleural infection, are associated with long hospital stay, high Intensive Care Unit admission rates, serious morbidity and mortality.

The cornerstone of treating pleural infection is the timely administration of appropriate antibiotics and adequate pleural fluid drainage. However, the initial choice of antibiotics is almost always empirical, and only able to be adjusted upon the return of positive microbiological results from the pleural fluid or relevant specimens. Indeed, the inadequate understanding of local bacteriology, drug resistance pattern and the lack of local guidance on empirical antibiotic coverage for pleural infection have led to heterogeneous prescription behaviour of antibiotics, which is a significant risk factor of the emergence of drug resistance (e.g., increasing incidence of methicillin-resistant Staphylococcus aureus), complications of antibiotic therapies (e.g., Clostridium difficile infections) and adverse patient outcomes.

Conventional culture of pleural effusion has been considered as an essential part of the treatment algorithm, but it carries two critical disadvantages that may limit clinical care:

1. Long turnaround time Although the Gram stain results can usually be available within one day, the final culture results and antimicrobial sensitivity pattern only return after up to five days. In case the pleural infection is not under control while waiting for the culture results, physicians would blindly escalate the coverage of antibiotics or proceed to additional pleural interventions, which may be avoidable if the microbiological results can be available early. The delayed return of drug sensitivity pattern may lead to the initial use of discordant antibiotics, which happened in 17 to 24% of patients in two local cohorts
2. Negative pleural fluid culture is common in pleural infection Up to 40% of pleural fluid in pleural infection failed to reveal any microorganism by conventional culture, due to a failure to culture fastidious organisms and antibiotics treatment before pleural fluid sampling. Our study group has found that 88% of patients with empyema received empirical antibiotics before any pleural intervention, which reflected the real-world practice. Although inoculating pleural fluid into a blood culture bottle may increase the culture yield in pleural infection by 20%, it is still far from satisfactory in clinical practice. In addition, the microbiological workup of non-respiratory specimens is frequently not informative. Thus, the use of antibiotics throughout the treatment course is purely empirical in those patients with culture-negative pleural infection. To overcome this problem and the often polymicrobial nature of the infections, very broad-spectrum agents are currently advocated. On the contrary, a lack of bacterial culture information with inadequate antibiotic coverage has been associated with an increased mortality rate.

Non-culture molecular studies have been applied to overcome the insufficiencies of conventional culture in the past decade, to characterize the spectrum of bacteria in pleural infection and the antibiotic resistance pattern within a short period of time and inform correct clinical decisions. Multiplex PCR (polymerase chain reaction) sequencing with a pneumonia panel is able to identify bacteria that are commonly involved in pneumonia. However, the aetiology of pleural infection is not completely a replicate of community-acquired pneumonia, and thus limit its utility. 16S ribosomal ribonucleic acid (rRNA) gene sequencing outperforms the conventional culture method by identifying more bacteria, especially anaerobes, in the pleural fluid. It provides a high-throughput and cost-efficient solution to screen the microbiota community for clinical samples with low bacterial biomass and/or high host genomic contamination. However, it has a limitation in detecting microbiome at the species level and may introduce PCR-biases that mask the true community composition. Dyrhovden R et al performed the first metagenomic sequencing analysis in Norway to understand the spectrum of bacteriology in confirmed empyema. They identified 385 bacterial detections, whereas culture detected 38 (10%) and 16S rRNA gene PCR/Sanger-based sequencing detected 87 (23%) in 64 patients with empyema. Such findings confirmed the disadvantage of the conventional culture method and broader coverage of antibiotics is required for the treatment of pleural infection. However, this metagenomic study was retrospective in nature, covered patients with empyema only and did not reflect the full spectrum of pleural infection. Chen et al employed a metagenomic sequencing study with functional analysis and identified two distinct microbiome clusters in pleural infection, Staphylococcus aureus as the core species (HA-SA type) and a more diverse microbial community (LA-SA type), which is different from the usual bacteriology pattern found in Hong Kong studies. They also identified the resistome of bacteria, including tetracycline and beta-lactam resistance, which is important to guide the appropriate antibiotic regimen in the initial treatment phase. The interpretation of these two studies was limited because the authors did not correlate the microbiological results with patient clinical outcomes. Recently, Kanellakis et al confirmed the predominantly polymicrobial nature of pleural infection by 16S rRNA metagenomic sequencing analysis, with a diverse bacterial spectrum. They also identified a distinct bacteriology spectrum in polymicrobial and monomicrobial cases and correlated the bacteriology with clinical outcomes.

Given the fact that geographical differences in bacteriology of pleural infection may exist and previous studies did not translate the modern molecular sequencing technologies in refining the initial antibiotic use, a local prospective study linking the laboratory and clinical findings using broader metagenomic sequencing methods is desired. We hypothesize that the 16S rRNA gene amplicon and shotgun metagenomic sequencing has a higher detection rate of bacteria in the pleural fluid and identification of antibiotic resistance patterns than conventional culture in patients with pleural infection. The results from the two molecular sequencing methods are complementary and can guide the initial antibiotic treatment, minimize the emergence of drug resistance, correlate the microbiological spectrum with the clinical outcomes and prepare for a large-scale territory-wide study by incorporating molecular technology into the management of pleural infection. The short turnaround time also makes this technology attractive to physicians and ultimately benefits the patient by improving the quality of care. As the diagnostic performance, technical requirements and cost of the two molecular sequencing methods are different, this study will provide fundamental information and guide their use in clinical settings.

The pathogenesis of community-acquired pleural infection is controversial. The conventional belief of transmigration of bacterial through the lung parenchyma to the pleural cavity was challenged by (1) different spectrum of causative pathogens for pneumonia and pleural infection, (2) the absence of radiological evidence of pneumonia, and (3) anaerobes are frequently found in pleural fluid culture but are uncommon in normal lung parenchyma because of high oxygen tension. Therefore, understanding the ultimate source of pathogens and understand their route of entering the pleural space is a critical step in preventing and treating pleural infection. Currently, there is very limited evidence associating the microbiota in pleural fluid of pleural infection, and those in the oral cavity and gastrointestinal tract.

The aim of the study is to characterize the full spectrum of causative bacteria in pleural infection and compare the diagnostic performance of identifying the pathogens between conventional culture, 16S rRNA gene amplicon and shotgun metagenomic sequencing. The antimicrobial resistance pattern of causative bacteria, clinical outcomes of pleural infection will also be compared between groups as categorized by the pattern of bacteriology identified by different methods.

Study group (patients with parapneumonic effusion)

Inclusion Criteria:

• Patients hospitalized for suspected pleural infection, irrespective of community-acquired or hospital-acquired in nature
• Pleural tapping will be performed for pleural fluid analysis
• Chinese ethnicity
• Aged 18 years old or above

Exclusion Criteria:

• Use of antibiotics for more than 24 hours for the current episode of infection
• On long-term local or systemic antibiotics
• Tuberculous pleuritis
• Pregnant or lactating women
• Failed to obtain informed consent due to patient's refusal or cognitive impairment

Control group (patients with effusions but not parapneumonic in nature)

Inclusion Criteria:

• Pleural effusion is not related to any infection including pneumonia (e.g. malignancy, fluid overload)
• Pleural tapping will be performed for pleural fluid analysis
• Chinese ethnicity
• Aged 18 years old or above

Exclusion Criteria:

• The use of systemic (including oral and intravenous) and inhaled antibiotics in the past 1 month
• On long-term local or systemic antibiotics
• Tuberculous pleuritis
• Pregnant or lactating women
• Failed to obtain informed consent due to patient's refusal or cognitive impairment

Study group: patients with parapneumonic effusion Control group: patients with pleural effusion but not parapneumonic in nature