Significance of Anaerobes and Oral Bacteria in Community-Acquired Pneumonia

Background Molecular biological modalities with better detection rates have been applied to identify the bacteria causing infectious diseases. Approximately 10–48% of bacterial pathogens causing community-acquired pneumonia are not identified using conventional cultivation methods. This study evaluated the bacteriological causes of community-acquired pneumonia using a cultivation-independent clone library analysis of the 16S ribosomal RNA gene of bronchoalveolar lavage specimens, and compared the results with those of conventional cultivation methods. Methods Patients with community-acquired pneumonia were enrolled based on their clinical and radiological findings. Bronchoalveolar lavage specimens were collected from pulmonary pathological lesions using bronchoscopy and evaluated by both a culture-independent molecular method and conventional cultivation methods. For the culture-independent molecular method, approximately 600 base pairs of 16S ribosomal RNA genes were amplified using polymerase chain reaction with universal primers, followed by the construction of clone libraries. The nucleotide sequences of 96 clones randomly chosen for each specimen were determined, and bacterial homology was searched. Conventional cultivation methods, including anaerobic cultures, were also performed using the same specimens. Results In addition to known common pathogens of community-acquired pneumonia [Streptococcus pneumoniae (18.8%), Haemophilus influenzae (18.8%), Mycoplasma pneumoniae (17.2%)], molecular analysis of specimens from 64 patients with community-acquired pneumonia showed relatively higher rates of anaerobes (15.6%) and oral bacteria (15.6%) than previous reports. Conclusion Our findings suggest that anaerobes and oral bacteria are more frequently detected in patients with community-acquired pneumonia than previously believed. It is possible that these bacteria may play more important roles in community-acquired pneumonia.


Introduction
Pneumonia is now the sixth and third leading cause of death in the United States and Japan, where 14.3/100,000 and 98.9/ 100,000 people die of the disease per year, respectively [1,2]. Pneumonia is also a leading cause of death in the elderly (.80 years old) in both countries [1,2]. It is estimated that the mortality of pneumonia will increase in aging population.
Having a precise understanding of the pathogens that cause pneumonia is very important to achieve prompt diagnoses and to determine proper antimicrobial treatments. However, according to previous reports, 10-48% of the causes of community-acquired pneumonia (CAP) were etiologically unknown when sputum and blood cultures were performed in combination with serological tests and tests for detecting urinary antigens [3][4][5][6][7][8][9]. In addition, relatively low incidences of anaerobes have been reported as causative bacteria (0-5.5%) [3][4][5][6][7][8][9]. It has been speculated that bacteria that are less culturable, such as anaerobes and oral bacterial flora, and they are assumed to be indigenous and tend to be ignored in sputum samples in ordinary clinical settings, may be responsible for the unknown bacteriological etiology in CAP. However, the incubation of the samples in agar plates under anaerobic conditions in clinical microbiology laboratories is not commonly performed.
Recently, the microbiota of the lower respiratory tract in patients with pulmonary infections, such as intensive care unit pneumonia [10], cystic fibrosis [11] and ventilator-associated pneumonia (VAP) [10], were studied using 16S ribosomal RNA (rRNA) gene amplification followed by clone library methods. In addition to identifying well-known causative pathogens of lower respiratory tract infections, these studies indicated the involvement of many bacteria that were previously thought to be nonpathogenic. Furthermore, information regarding bacteria obtained in the past several years using new molecular biology techniques (16S quantitative PCR followed by pyrosequencing) has highlighted the existence and possible clinical roles of the microbiota of the lower respiratory tract in patients with chronic obstructive pulmonary disease [12][13][14] and even in healthy subjects [15].
We previously reported the diagnostic utility of a clone library analysis of the 16S rRNA gene using bronchoalveolar lavage (BAL) fluid for bacteriological information in patients with pneumonia caused by Legionella sp. [16] and Leptotrichia sp. [17]. This molecular method can detect the phylotypes whose 16S rRNA gene sequences are the most similar to those of the type strains, and can determine the ratio of phylotypes (bacterial flora) in each specimen in a cultivation-free fashion.
The diagnostic utility of the culture of BAL fluid using fiberoptic bronchoscopy with higher detection rates than sputum samples in CAP patients was also reported [18].
In the present study, we performed bronchoscopy to evaluate the causative pathogens in CAP patients, and BAL specimens were analyzed by both a microfloral molecular analysis of the 16S rRNA gene and ordinary cultivation methods, in combination with serological assays and detection of urinary antigens.

Subjects
Sixty-four consecutive CAP patients in our university hospital and referred hospitals between April 2010 and December 2011 were enrolled in this study. Bronchoscopy was performed to evaluate the causative pathogens in the lesions of these pneumonia patients. CAP was defined according to the Infectious Diseases Society of America (IDSA)/American Thoracic Society (ATS) guidelines for diagnosing CAP in adults [19]. This study excluded patients with healthcare-associated pneumonia (HCAP) and hospital-acquired pneumonia (HAP) [20].This study was approved by the Human and Animal Ethics Review Committee of the University of Occupational and Environmental Health, Japan (No.09-118). Written informed consent was obtained from either the patients or their guardians. If the patients were under 20 years old, their parents provided written informed consent on their behalf. The following patient information was collected: age, sex,  underlying diseases, clinical manifestations, and laboratory and radiological findings. BAL specimens obtained from 30 patients with idiopathic interstitial pneumonias (IIPs) using the same methods were also evaluated as representative samples of noninfectious pulmonary diseases.

Sample Collection
Fiberoptic bronchoscopy was performed according to the British Thoracic Society guidelines for diagnostic flexible bronchoscopy [21]. Gargling with povidone iodine solution was performed before bronchoscopy to minimize contamination by oral bacteria, and a fiberoptic bronchoscope was then introduced transorally into the trachea by passing it through the vocal cords without any contacts or aspiration to avoid oral bacterial contamination. BAL specimens were then obtained from the affected lesions using 40 ml of sterile saline. Moreover, sputum samples were also evaluated in patients with sputum production.

Total Bacterial Cell Counts and Cell Lysis Efficiency Analyses
To provide a precise evaluation of the microbiota, we evaluated the total bacterial cell counts and the efficiency of cell lysis using epifluorescent microscopy, as previously reported [22].

Microbiological Examination
The BAL specimens and the sputum samples were quantitatively cultivated under aerobic and anaerobic conditions as described previously [22].
Serological methods using single or paired sera were used to examine the presence of antibodies against Mycoplasma pneumoniae Complement Fixation Antigen (Denka Seiken, Tokyo, Japan), and Chlamydophila psittaci Complement Fixation Antigen (Denka Seiken, Tokyo, Japan). The level of anti-Chlamydophila pneumoniae antibodies was determined by the SeroCP ELISA for immunoglobulin G (IgG) and IgA (Savyon and Hain Lifescience, Nehren, Germany). Urinary antigen tests to detect Streptococcus pneumoniae and Legionella pneumophila (Binax, Portland, ME, USA) were also performed.

Criteria for a Conventional Etiologic Diagnosis
Bacteria were considered to be causative organisms when they were isolated from blood cultures. Any microorganism isolated from the BAL specimens was considered to be a presumptive pathogen when its concentration reached $10 4 colony-forming units (CFU)/ml in the quantitative cultures [23,24].
For serological assessment of M. pneumoniae and C. pneumoniae, a four-fold increase in antibody titer levels between the paired sera was considered to be presumptive. L. pneumophila and S. pneumoniae were considered to be presumptive agents when the urinary antigen tests were positive.

DNA Extraction
DNA samples were extracted from the BAL specimens by vigorously shaking them with sodium dodecyl sulfate (final concentration: 3.0%) and glass beads, as reported previously [22].

PCR Conditions
16S rRNA genes were amplified with a GeneAmp PCR system 9700 thermocycler (Applied Biosystems; Foster City, CA). The reaction mixtures containing the universal primers set [25] (E341F; 59-CCTACGGGAGGCAGCAG-39 and E907R; 59-CCGTCAATTCMTTTRAGTTT-39) and AmpliTaq Gold DNA polymerase LD (Applied Biosystems; Foster City, CA) were incubated in a thermocycler at 96uC for 5 min. This was followed by 30 cycles at 96uC for 30 s, 53uC for 30 s and 72uC for 1 min and a final elongation step at 72uC for 7 min.

Clone Library Construction and Determination of Nucleotide Sequences
The PCR products were cloned with a TOPO TA cloning kit (Invitrogen; Carlsbad, CA) according to the manufacturer's instructions. A total of 96 colonies were randomly selected from each clone library for sequencing analysis. The partial fragments of the cloning vectors (pCR II) containing inserted PCR products were amplified with AmpliTaq Gold DNA polymerase and a primer set (M13Forward; 59-GTAAAACGACGGCCAG-39 and M13Reverse; 59-CAGGAAACAGCTATGAC-39). After the primers and deoxyribonucleotide triphosphate were eliminated from the PCR mixture with an ExoSAP-IT (GE Health care UK Ltd.; England, UK) according to the manufacturer's instructions, a 1-ml aliquot was used as a template for the sequencing reaction. The sequencing reactions were accomplished with primers ''M13Forward'' and the BigDye Terminator Cycle Sequencing Kit v3.1 (Applied Biosystems). The nucleic acid sequences were determined on a 31306l Genetic Analyzer (Applied Biosystems).

Homology Searching
Highly accurate sequences selected by the Phred quality values were compared with the 16S rRNA gene sequences of the type strains using the basic local alignment search tool (BLAST) algorithm, as described previously [22]. A phylotype sharing 97% or higher homology with the sequence of the type strain was assumed to be a presumptive species, as described previously [26], and a phylotype with a sequence sharing between 90% and 97% of the type strain was assumed to be a presumptive genus in the present study.

Assessment of Pneumonia Severity and 30-day Mortality
The assessment of the severity of pneumonia in each patient was conducted using the pneumonia severity index (PSI) [27]. The mortality 30 days after admission was also evaluated.

Total Bacterial Numbers Obtained with Epifluorescent Microscopic Evaluations and Cell Lysis Efficiency Analyses
The number of bacteria in each BAL specimen was counted using an epifluorescent microscopic analysis. The numbers of bacteria ranged from 1.3610 4 to 3.7610 9 (median 2.5610 6 ) cells/ ml ( Table 2). The efficiency of cell lysis was maintained at 80% or greater in all samples.
Neither pathogenic organisms nor antigens were detected in 12 (18.8%) out of the 64 CAP patients using the conventional methods, including cultivation, serological examination, and urinary antigen detection. In contrast, anaerobes (Prevotella spp. in four; Fusobacterium spp. in three; Clostridium spp. in one) were highly detected in eight (66.7%) out of these 12 patients by the molecular method.
For descriptive purposes, we defined the ''monobacterial dominant group'' as the group which included patients in whom the first dominant phylotype comprised over 80% of the detected bacteria, while the other patients were assigned to the ''mixedbacterial group.'' According to this definition, 33 patients were categorized as belonging to the monobacterial group, and 31 patients were defined as belonging to the mixed-bacterial group (Figure 2).
The BAL specimens obtained from 30 patients with IIPs as representative of noninfectious pulmonary diseases were also evaluated. These specimens showed no bacterial growth and no 16S rRNA gene amplification in any of the subjects using either conventional cultivation methods or the molecular method. In addition, epifluorescent microscopic analyses of these BAL specimens also showed results below the limit of detection (,1.3610 4 cells/ml) in all patients with IIPs, in comparison to the results from 64 CAP patients (from 1.3610 4 to 3.7610 9 ; median 2.5610 6 cells/ml). (Figure 3). Table 3 shows the ratio of detected bacterial species in each PSI category and each age group. In mild cases, M. pneumoniae was the most dominant species, followed by anaerobes, H. influenzae and S. pneumoniae. S. pneumoniae, M. catarrhalis, and oral streptococci were predominantly detected in severe cases.
As the first dominant phylotype, M. pneumoniae was the most frequently detected phylotype in patients ,40 years of age, and M. pneumoniae, H. influenzae and anaerobes were primarily detected in patients aged 40-64 years. S. pneumoniae, H. influenzae, M. catarrhalis, and oral streptococci were predominantly detected in patients .64 years of age.

Discussion
In the present study, we analyzed BAL specimens obtained from 64 CAP patients using a clone library analysis of the 16S rRNA gene. To the best of our knowledge, this is the first report to show a higher incidence of anaerobes in CAP patients than was previously believed. Previous reports [3][4][5]9,28,29] have shown that 10-48% microorganisms are etiologically unidentified using traditional culture methods for sputum, in combination with serological and/ or specific urinary antigen detection in CAP patients. In contrast to the former reports using cultivation methods, this molecular method could detect bacterial phylotypes in all CAP patients, and the predominant phylotypes were similar to those found in previous reports using traditional cultivation methods [3][4][5]. (Figure 1 and Table 2) In addition, well-known common pathogens of CAP were primarily detected in the ''monobacterial dominant group,'' whereas most patients in the ''mixed-bacterial group'' showed multiple bacterial phylotypes including anaerobes and/or oral streptococci.
Using this molecular method, obligate anaerobes such as Prevotella spp. and Fusobacterium spp. (10/64, 15.6%) and oral streptococci, including S. intermedius (6/64, 9.4%), were preferentially detected, especially in CAP patients with unknown etiologies indicated by cultivation-based methods. These results suggest that resident oral streptococci and anaerobes might be the primary bacteria responsible for the unknown causative pathogens of CAP in the previous reports [3][4][5]. These bacteria are generally not considered to be causative pathogens of CAP; however, the higher detection rates of anaerobes in the CAP patients in this study than was previously believed to be present suggest that these bacteria may play important roles in CAP.
We previously applied this molecular method for the etiological evaluation of bacterial pleurisy, and reported that anaerobes were detectable in approximately half of bacterial pleurisies [22]. Moreover, recent molecular studies have shown that anaerobes are frequently detected in patients with stable cystic fibrosis [11].
Oral bacteria (six non-pneumococcal streptococci, three Neisseria spp. and one Corynebacterium sp.) were detected as the first dominant phylotypes in 10 CAP patients. S. intermedius is a member of the S. anginosus group, and these bacteria have been reported to range from 1.1% to 3.1% [3][4][5] as causative bacteria in CAP patients. Only a few reports of CAP caused by S. viridans (or oral streptococci) or Neisseria spp. have been reported [30], although Lambotte et al. reported that oral streptococci and Neisseria spp. could be causative bacteria in VAP patients [31]. Moreover, 76 (6.8%) out of the 1118 CAP patients showed bacteremic pneumonia and seven (9.2%) of these had positive blood cultures for various non-pneumococcal streptococci in a previous study [32]. All patients with Neisseria spp. were relatively immunocompromised in this study; therefore, oral bacteria may preferentially cause CAP in relatively immunocompromised hosts. Epifluorescent microscopic evaluations in CAP patients and patients with IIPs ( Figure 3) demonstrated that the combination of this molecular method and epifluorescent microscopic evaluation detected some bacterial phylotypes only in bacterial infectious diseases. Using this method together with the bronchoscopic method, we were able to avoid any contamination with oral bacteria, which may make it possible to distinguish lower respiratory tract bacterial infections from other noninfectious bronchopulmonary diseases. There are several limitations associated with this study that should be kept in mind when interpreting the results. First, the universal primers we used could not amplify all of the bacterial 16S rRNA genes, and the sensitivity of the primers was approximately 92% for the bacterial species registered in the Ribosomal Database Project II database. However, the remaining approximately 8% of the bacteria undetectable using these primers does not include any reported human pathogens. Second, the number of clones analyzed in this study was approximately 100 per library, suggesting that this method may not be able to detect bacterial 16S rRNA gene sequences when they are present at very small fractions (less than 1% of each sample). The sequencing depth used in this study is not suitable to detect Mycobacterium tuberculosis, which is an important bacterium to assess when obtaining a diagnosis of respiratory disease, even if the bacterium is a minor constituent of clinical specimens.

Conclusions
We evaluated the causative bacterial species in CAP patients using a microfloral analysis as a cultivation-independent method to detect the presence of the 16S rRNA gene in BAL specimens. The results of our study demonstrate that the incidence of anaerobes and oral bacteria in CAP patients, especially in patients with mild PSI, is higher than previously reported. Therefore, clinicians should consider that anaerobes and oral bacteria are more frequent pathogens than previously believed in CAP patients.