Use of the Proportion of Neutrophils in Bronchoalveolar Lavage Fluid for Rapid Diagnosis of Bacterial Pneumonia in Children

Introduction

Pneumonia is a prevalent disease in children, and a leading cause of death among those under 5 years old. The most common pathogens responsible for childhood pneumonia are viruses, bacteria, and bacteria causing atypical pneumoni, including mycoplasma and chlamydia.^1,2 The pathogens of childhood pneumonia are very diverse and often difficult to predict. The lack of timely and accurate etiological diagnosis could substantially affect childhood pneumonia outcomes.³ Rapid identification of the type of pneumonia pathogen can facilitate early treatment with precise medication, alleviate suffering in paediatric patients, reduce medical resource waste, and prevent antibiotic misuse. However, differentiating between bacterial and non-bacterial pneumonia in children remains a significant challenge. Currently, common markers for bacterial infection include peripheral blood white blood cell count (WBC), neutrophil percentage (NE%), C-reactive protein (CRP), and procalcitonin (PCT). These indicators can help differentiate bacterial and non-bacterial pneumonia in children; however, their diagnostic sensitivity and specificity are influenced by various factors. For example, these markers can be elevated because of non-infectious inflammatory responses and organ damage, complicating the use of WBC, NE%, CRP, and PCT for differentiating viral from bacterial pneumonia.⁴ Although antibody detection for respiratory pathogens can facilitate the diagnosis of some bacterial and Mycoplasma pneumoniae infections, this method is hampered by high false-negative and false-positive rates. Therefore, its diagnostic value for paediatric pneumonia is limited.⁵ Sputum bacterial culture is also suboptimal because of its time-consuming nature and relatively low sensitivity. Furthermore, obtaining sputum samples for bacterial culture is particularly challenging in infants and young children, and culturing can easily fail because of poor sample quality. While Polymerase chain reaction is diagnostically valuable for detecting common respiratory pathogens,⁶ it is limited to predefined microorganisms, with fair specificity and high cost. High-throughput sequencing of sputum, throat swabs, and bronchoalveolar lavage fluid (BALF) has great potential in pathogen differentiation in paediatric pneumonia;⁷ however, its lengthy processing time and high costs make rapid diagnosis impractical.

BALF cytological analysis has been increasingly used in clinical practice. This method is critical in diagnosis, monitoring treatment efficacy, and predicting outcomes for respiratory diseases, particularly those affecting the lower respiratory tract.^8–10 It involves rapid preparation and observation of BALF smears to infer the etiological type of infection based on inflammatory cell profiles, offering speed and convenience for aetiological diagnosis. In healthy individuals, most BALF cells are macrophages, and the neutrophil percentage in BALF (BALF-NE%) is very low. However, BALF-NE% is elevated in bacterial lung infections, making this parameter crucial for diagnosing infectious lung diseases in adults, although the optimal threshold can vary significantly.^11,12 Despite its established use in adults, quantitative BALF-NE% thresholds specifically validated for distinguishing bacterial from viral pneumonia in children remain inadequately defined. A systematic search of PubMed and Web of Science using keywords [(“bronchoalveolar lavage fluid” OR “BALF”) AND (“neutrophil percentage” OR “neutrophil proportion” OR “cell differential”) AND (“pneumonia” OR “lower respiratory tract infection”) AND (“child*” OR “pediatric” OR “paediatric”)] identified limited pediatric data. Furthermore, the differentiation between viral and bacterial pneumonia is more difficult in children than in adults using measurements of WBC, CRP, PCT, or chest imaging. Therefore, in this study, we aimed to evaluate the effectiveness of BALF cytological analysis in rapidly differentiating between bacterial and non-bacterial pneumonia in paediatric patients and to establish the optimal threshold of BALF-NE% for diagnosing bacterial pneumonia in children.

Patients and Methods

Participants

The study was conducted at the Affiliated Hospital of Guangdong Medical University, located at the southernmost tip of mainland China with a tropical climate. Patients who underwent bronchial examination at its Children’s Medical Centre between January 2019 and December 2022 were selected as study participants, and those clinically diagnosed with laryngomalacia served as the control group. These patients underwent bronchoscopy for evaluation of inspiratory stridor but had no lower respiratory tract infection or lung parenchymal inflammation, thereby providing BALF samples with normal pulmonary cytology for comparison.

The inclusion criteria for this study were as follows: (1) age: >29 days and < 14 years; (2) clinically diagnosed with either bacterial or viral pneumonia, having undergone bronchoalveolar lavage (BAL) with subsequent BALF smear and BALF cytological analysis, and with the results of BALF culture or metagenomic next-generation sequencing (mNGS) confirming the presence of the pathogen. BALF samples were inoculated onto blood agar, chocolate agar, and MacConkey agar and incubated at 35–37°C in 5% CO₂ for 18–48 hours. This conventional culture method identified typical bacterial pathogens including Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus aureus, and other common respiratory bacteria. Fastidious atypical pathogens (Chlamydia pneumoniae, Mycoplasma pneumoniae, and Bordetella pertussis) were not cultivable by this method and were identified exclusively by mNGS.

Further, participants who met the following criteria were excluded: (1) BALF mNGS identified both viral and bacterial pathogens, with a clinical diagnosis of mixed viral and bacterial lung infection; (2) presence of bloodstream infection, urinary tract infection, or bacterial infections at other sites; and (3) inadequate quality of the BALF sample,¹³ ie, squamous epithelium ≥5% or columnar epithelium ≥5%, BALF recovery rate <30%, or contamination with blood.

Grouping and Diagnostic Criteria

The diagnostic criteria for the patients included in the bacterial pneumonia group⁴ were as follows: presence of symptoms such as fever, cough, or shortness of breath; wet rales detected on lung auscultation and signs of pneumonia in pulmonary imaging, with BALF culture or BALF mNGS testing positive for bacteria. The bacterial pneumonia group was categorised into severe and non-severe pneumonia subgroups based on their disease severity. Individuals who met at least one of the following diagnostic criteria were included in the severe pneumonia subgroup:⁴ shortness of breath (infants with a respiratory rate >70 times/min, toddlers and older children with a respiratory rate >50 times/min); cyanosis or depressions in the suprasternal fossa, supraclavicular fossa, and intercostal spaces; an oxygen saturation of ≤92% when not inhaling oxygen, disturbed consciousness, or the presence of extrapulmonary complications.

The diagnostic criteria for patients in the viral pneumonia group were as follows:¹⁴ symptoms such as fever, cough, or shortness of breath, chest imaging indicative of interstitial pneumonia, and detection of respiratory viruses by BALF mNGS.

The diagnostic criteria for patients in the laryngomalacia group were as follows:¹⁵ clinical manifestations of inspiratory stridor and upper airway obstruction, with endoscopic features characteristic of laryngomalacia.

Study Methods

Biochemical and Cytological Analysis

The clinical data collected included age and sex, symptoms and signs; results of laboratory tests including peripheral blood WBC, NE%, CRP, PCT, and BALF cytological analysis; and imaging results of chest computed tomography or radiography. Regarding the collection, processing, and cell counting of BALF samples, BALF was collected according to the standard procedure recommended by the guidelines.¹³ The initial BALF aliquot was discarded to avoid oropharyngeal contamination. Subsequent aliquots were allocated to mNGS/microbial culture (second) and cytological analysis (third).¹⁶ Isolated bacterial colonies were identified using the bioMérieux Vitek2 COMPACT automated system with Vitek2 GP (Gram-positive) and GN (Gram-negative) identification cards (bioMérieux, France), following manufacturer protocols. Antimicrobial susceptibility testing was performed concurrently where clinically indicated. The mNGS detection employed high-throughput sequencing technology to analyse microbial nucleic acid sequences in specimens, followed by comparative identification against existing microbial sequence databases. The total cell count in the BALF was quantified using a counting chamber, and presented as 1×10⁶/mL. In instances of excessively high cell counts, dilution was performed to adjust the concentration to 5×10⁶/mL, and the tubes were stored on ice for subsequent use. Cell counting of BALF samples was performed using an improved Neubauer counting chamber. Slides were initially observed at lower magnification (<40×) using a light microscope for an overall examination, followed by cell classification and counting at higher magnification (100×). Two slides were counted, with at least 600 cells per slide, and the average of the two counts was taken as the final result. Figure 1 shows a representative BALF sample under an oil immersion lens at 100× magnification.

Figure 1 Cell classification in BALF samples under an oil immersion lens at 100× magnification. (a) Bacterial pneumonia, arrow indicating neutrophil (NE); (b) Control group (laryngomalacia), arrow indicating macrophage; (c) Unqualified BALF sample, arrow indicating ciliated columnar epithelial cell.

Molecular Analysis

mNGS was performed by DaAn Gene (Guangzhou, China). Specimens were collected in sterile containers, transported immediately, and processed according to standard laboratory protocols. Total nucleic acid was extracted using the DA1780 kit (DaAn Gene), libraries were constructed using the DYQ-N0009 kit, and sequencing was performed on the MGI-200 sequencer after Qubit-based quality control. Negative and positive controls were included in each batch.

Sequencing data were processed using fastp for adapter trimming and quality filtering, followed by host sequence removal (Bowtie2, hg38). Microbial alignment was performed using SNAP-aligner against a customized database (~31,000 genomes) constructed following NCBI standards,¹⁷ with pathogen selection based on Johns Hopkins ABX Guide criteria. A pathogen was considered positive when ≥3 non-overlapping regions of the genome were covered.

Statistical Analysis

All data analyses were performed using SPSS 25.0 statistical software. Quantitative data following normal distribution were presented as and between-group comparisons were conducted using t-tests or analysis of variance. Data with skewed distribution were presented as median (interquartile range) [M (IQR)], and between-group comparisons were performed using non-parametric tests. Count data comparisons were conducted using the chi-square test. The diagnostic efficacy of BALF-NE% for bacterial pneumonia was assessed with receiver operating characteristics (ROC) curve analysis, with P<0.05 considered statistically significant.

Results

Basic Characteristics of the Patients

Overall, 202 patients who met the diagnostic criteria for bacterial or viral pneumonia, underwent bronchoscopic examination, and had BALF culture or BALF mNGS confirming the presence of the pathogen, were identified. After excluding 30 cases with unqualified BALF samples and 31 cases where BALF mNGS detected both viral and bacterial pathogens, 141 patients with pneumonia were ultimately included, of which 68 were classified as severe cases. Among these, 116 cases were diagnosed with bacterial pneumonia (60 severe cases) and 25 with viral pneumonia (eight severe cases) (Figure 2). Furthermore, 13 cases diagnosed with laryngomalacia via bronchoscopy were also included as the normal control group.

Figure 2 Study flowchart of patient inclusion and exclusion criteria. Abbreviations: BALF, bronchoalveolar lavage fluid; mNGS, metagenomic next-generation sequencing.

No statistically significant differences were observed among the three groups in regard to sex, age, cough symptoms, presence of dry or wet lung sounds, or the proportion of patients with wet rales (p>0.05). Laboratory test results revealed that the proportion of alveolar macrophages (BALF-AM%) in the bacterial pneumonia group was significantly lower than in the viral pneumonia and control groups (p<0.05). The bacterial pneumonia group had higher BALF-NE%, WBC, NE%, CRP, and PCT levels than those of the viral pneumonia and control groups (all p<0.05) (Table 1).

Table 1 Baseline Characteristics and BALF Composition by Study Group

This study included a total of 116 cases of bacterial pneumonia, with 48 identified through BALF bacterial culture and 68 through mNGS. Streptococcus pneumoniae was identified in 38 cases, with 23 (60.5%) in infants and 15 (39.5%) of severe pneumonia. Mycoplasma pneumoniae was found in 16 cases, predominantly in children of 6–14 years old, accounting for 10 cases (62.5%), with three (18.8%) case of severe pneumonia. Chlamydia trachomatis was identified in 13 cases, exclusively in infants (100%), with nine (69.2%) classified as severe pneumonia. Haemophilus influenzae was detected in 11 cases, with 10 (90.0%) in infants and seven (63.6%) classified as severe pneumonia. Bordetella pertussis was identified in 10 cases, exclusively in infants (100%), with seven (70.0%) classified as severe pneumonia. Staphylococcus aureus was detected in nine cases, with six (66.7%) in infants and six (66.7%) classified as severe pneumonia. Other bacteria were responsible for 19 cases, with nine (47.4%) in infants and 13 (68.4%) classified as severe pneumonia. Figure 3 provides details of the proportion of severe cases of bacterial pneumonia.

Figure 3 Proportion of severe cases by bacterial pathogen. Red indicates severe pneumonia; blue indicates non-severe pneumonia. Streptococcus pneumoniae was the most common pathogen (38 cases, 32.8%); Bordetella pertussis showed the highest severity rate (70.0%, 7/10 severe).

Overall, 25 cases of viral pneumonia were included in this study, all of which were identified through BALF mNGS. Among these, 16 were infections with a single virus, and nine were mixed virus infections. The viruses identified included respiratory syncytial virus (12 strains), adenovirus (seven strains), human metapneumovirus (four strains), parainfluenza virus (four strains), bocavirus (three strains), rhinovirus (two strains), influenza A virus (one strain), and influenza B virus (one strain).

Laboratory Indicators for Different Types of Pneumonia

Figure 4 demonstrates the BALF-NE% for each patient within the bacterial pneumonia, viral pneumonia, and control groups, and further presents the overall differences. Generally, BALF-NE% exhibited the highest values in the bacterial pneumonia group, followed by the viral pneumonia group, and was the lowest in the control group (laryngomalacia). Notably, the difference in BALF-NE% was more pronounced between severe cases of bacterial and viral pneumonia (Figure 5).

Figure 4 BALF-NE% in bacterial pneumonia, viral pneumonia, and control groups. The bacterial pneumonia group showed a significantly higher BALF-NE% than that in the viral pneumonia group. Specifically, all samples (13/13) had a BALF-NE% ≤10% in the normal control group, 92.0% (23/25) had a BALF-NE% ≤28% in the viral pneumonia group, and 93.1% (108/116) had a BALF-NE% >28% in the bacterial pneumonia group. Abbreviations: BALF-NE%, neutrophil percentage in bronchoalveolar lavage fluid. Notes: **p<0.001, and ***p<0.0001.

Figure 5 Difference in BALF-NE% between severe cases of the bacterial pneumonia and viral pneumonia groups. The bacterial pneumonia group showed a significantly higher BALF-NE% that in the viral pneumonia group; 87.5% (7/8) of cases in the viral pneumonia group had a BALF-NE% ≤28%, and all samples (60/60) in the bacterial pneumonia group had a BALF-NE% >28%. Abbreviation: BALF-NE%, neutrophil percentage in bronchoalveolar lavage fluid. Note: ***p<0.0001.

The bacterial pneumonia group was further classified into subgroups based on the specific pathogen as follows: bacteria causing typical pneumonia, Mycoplasma pneumoniae, Chlamydia trachomatis, and Bordetella pertussis. The median BALF-NE% in the Mycoplasma pneumoniae, Chlamydia trachomatis, and Bordetella pertussis subgroups was significantly higher than that of bacteria causing typical pneumonia (all p<0.05) (Figure 6).

Figure 6 BALF-NE% in bacterial pneumonia subgroups: typical versus atypical pathogens. BALF-NE% in Mycoplasma pneumoniae, Chlamydia trachomatis, and Bordetella pertussis subgroups was significantly higher than that of typical bacterial pneumonia; 92.3% (12/13) of the Chlamydia trachomatis subgroup had a BALF-NE% >70%, and the proportion was 100% (16/16) for the Mycoplasma pneumoniae subgroup, 80.0% (8/10) for the Bordetella pertussis subgroup, and 45.5% (35/77) for the typical bacteria subgroup. Abbreviations: BALF-NE%, neutrophil percentage in bronchoalveolar lavage fluid. Note: *p<0.05, **p<0.001.

ROC Curve Analysis of Various Laboratory Indicators in Diagnosing Bacterial Pneumonia

ROC curves were plotted to assess the diagnostic performance of BALF-NE%, CRP, PCT, peripheral blood WBC, and peripheral blood NE% for bacterial indicators for differentiating bacterial pneumonia from viral pneumonia (Table 2).

Table 2 Diagnostic Performance of Laboratory Indicators for Bacterial Pneumonia: ROC Curve Analysis

The BALF-NE% had the highest area under the receiver operating characteristics curve at 0.986 (95% confidence interval: 0.963–1.000, standard error: 0.012) (p<0.0001) for diagnosing bacterial pneumonia. When setting the threshold of BALF-NE% >28%, the sensitivity and specificity were 98.8% and 94.1%, respectively (Figure 7).

Figure 7 ROC curves for BALF-NE% in diagnosing bacterial pneumonia. BALF-NE% had the highest area under the receiver operating characteristics curve at 0.986 (95% CI: 0.963–1.000, SE 0.012) (p<0.0001) for diagnosing bacterial pneumonia. Abbreviations: ROC, receiver operating characteristic; BALF-NE%, neutrophil percentage in bronchoalveolar lavage fluid, CI, confidence interval, SE, standard error, CRP, C-reactive protein, PCT, procalcitonin, WBC, white blood cell.

BALF-NE% in Bacterial Pneumonia by Severity

Overall, the severe pneumonia group had higher BALF-NE% than the non-severe group (p<0.05). In the subgroup of bacteria causing atypical pneumonia (atypical bacteria subgroup), the BALF-NE% was higher in the severe group than in the non-severe group (p<0.05). However, no significant difference was observed in BALF-NE% between severe and non-severe pneumonia cases caused by virus or atypical bacteria (Mycoplasma pneumoniae and Chlamydia trachomatis) (p>0.05) (Table 3).

Table 3 BALF-NE% in Bacterial Pneumonia Cases Stratified by Severity

Discussion

This study demonstrates that BALF-NE% is a highly accurate biomarker for rapid differentiation of bacterial from viral pneumonia in children, with diagnostic performance superior to conventional serum inflammatory markers. The established threshold of >28% enables reliable identification of bacterial etiology, facilitating early targeted antibiotic therapy and reducing empirical treatment delays in pediatric patients.

The selection criteria for patients in our study were quite stringent. In addition to receiving a clinical diagnosis of pneumonia, all participants underwent aetiological diagnosis through BALF testing, and potential cases of mixed viral and bacterial pneumonia were excluded. Considering that bronchoscopy is generally reserved for hospitalised pneumonia patients with severe or complicated conditions, our cohort primarily comprised complex and severe cases. Bacterial infection is one of the leading causes of severe pneumonia in children, with infants and toddlers being particularly susceptible to bacterial pneumonia. The mortality rate for severe bacterial pneumonia is high, while the rapid identification of the pathogen is critical for treatment, and the rational use of antibiotics is pivotal to reducing mortality rates.^4,18,19 Our study encompassed a substantial proportion of severe cases, covering pathogens causing typical pneumonia, atypical pneumonia, and viral pneumonia, providing a robust representation of pneumonia cases.

Our findings indicate that while the clinical diagnosis of pneumonia can be readily established based on factors such as the patient’s age, the manifestation of fever and cough, lung auscultation sounds, and imaging findings, distinguishing between bacterial and viral pneumonia or identifying the specific bacterial infection type based solely on these factors remains challenging. Although serum CRP, PCT, and peripheral blood WBC and NE% levels are usually higher in patients with bacterial pneumonia than in those with viral pneumonia, these markers still suffer from suboptimal sensitivity and specificity. Many patients with bacterial pneumonia do not necessarily have higher levels of PCT, CRP, or peripheral blood NE%.^20,21 Furthermore, unlike adult pneumonia, where clinical-radiological differentiation is relatively straightforward, children present with nonspecific symptoms and rapid disease progression, necessitating rapid diagnostic tools. BALF-NE% assessment meets this need by providing etiological clues within hours, without requiring sophisticated equipment.

Bronchoscopy is extensively used in clinical practice, and BALF is critical in diagnosing lower respiratory tract infections. Cytological examination involving the counting of inflammatory cells in BALF facilitates the assessment of lung infection and inflammation. Its ease of operation, speed, and repeatability contribute to its application in clinical practice. In cases of bacterial infection of the lungs, a substantial increase in neutrophils occurs within 4 hours of infection, accompanied by significant neutrophil exudation into the alveoli. This process is crucial for the immune response during the early stages of bacterial pneumonia.²²

This study discovered that BALF-NE% was extremely low in healthy individuals; however, it was significantly increased in patients with bacterial pneumonia, and was significantly higher than in those with viral pneumonia. Patients with viral pneumonia also displayed a slightly elevated BALF-NE% compared to healthy individuals, establishing a clear differentiation in BALF-NE% levels among these three groups. Setting the threshold of BALF-NE% >28%, the sensitivity and specificity for distinguishing bacterial from viral pneumonia were 98.8% and 94.1%, respectively. Notably, this study revealed that, amongst cases of severe pneumonia, all had BALF-NE% >28%. This sensitivity and specificity surpassed those of conventional inflammatory markers, highlighting the remarkable diagnostic capability of BALF-NE% in identifying bacterial pneumonia in children and potentially surpassing its performance in adults.¹³ The high diagnostic accuracy (AUC 0.986) and immediate clinical applicability of BALF-NE% assessment support its potential utility. Nevertheless, prospective multicenter validation is warranted to confirm the generalizability of the >28% threshold before widespread clinical adoption. We are currently planning such a study in diverse clinical settings.

This study demonstrated that atypical bacterial infections, whether caused by Chlamydia trachomatis, Mycoplasma pneumoniae, or Bordetella pertussis, exhibited elevated BALF-NE% levels, with the majority of BALF-NE% values exceeding 70%. This effectively distinguishes atypical bacterial infections from typical bacterial infections. Additionally, Bordetella pertussis and Haemophilus influenzae caused substantial severe disease in our cohort (70.0% and 63.6%, respectively), despite being vaccine-preventable While pertussis vaccination is mandatory in China, Haemophilus influenzae type b vaccine remains optional with variable coverage. Our findings support strengthening Hib immunization and pertussis booster strategies, with BALF-NE% serving as a complementary tool for rapid outbreak detection.

Neutrophils are involved in the onset and development of lung lesions induced by bacterial infections. Upon congregating at the infection site, neutrophils engulf bacteria and degranulate during phagocytosis, releasing lysozyme and other lysosomal enzymes to kill the bacteria. Additionally, neutrophils secrete an array of inflammatory factors to regulate the activity of other immune cells, such as promoting the activation and differentiation of AM and lymphocytes.²³ Neutrophils are also critical in constraining bacterial spread in the lungs and clearing the infection. This study showed a significant reduction in BALF-AM% in patients with bacterial pneumonia compared to the control group, indicating that nonspecific immunity may be compromised in these patients.²⁴ Furthermore, this study demonstrated an increase in BALF-NE% and a decrease in BALF-AM% in patients with severe bacterial pneumonia. This phenomenon could be attributed to the bacterial uptake of AMs surpassing their capacity during the initial phases of infection, leading to cell death or functional impairment of AMs. Additionally, bacterial endotoxins can potentially damage or induce apoptosis in AMs, subsequently reducing AM numbers.²⁵

Additionally, our study revealed that patients with severe pneumonia typically have higher BALF-NE% than those with non-severe pneumonia, potentially owing to the consistent increase in neutrophils from high levels of infection and immune imbalance in severe bacterial pneumonia. Consequently, this excessive neutrophil activity could amplify inflammatory responses and tissue damage, further exacerbating the patient’s condition. The release of proteases and various inflammatory factors by neutrophils is also associated with increased disease severity and mortality.²⁶ Therefore, BALF-NE% can be valuable in assessing the severity of bacterial pneumonia.

This study has certain inherent limitations. The study is a single-centre study with a limited cohort size, and factors like prevailing pathogens and clinical practices can vary geographically. In addition, cohort primarily includes hospitalized children with moderate to severe pneumonia who underwent bronchoscopy. As such, the findings may not be readily generalizable to outpatient populations or those with milder disease presentations. Although we excluded confirmed cases of mixed viral and bacterial pneumonia, some included cases of bacterial pneumonia were diagnosed through BALF bacterial culture, and the possibility of viral pneumonia in the early stages of the disease in these patients exists. Nevertheless, we conducted comprehensive evaluations involving epidemiology, symptoms, and imaging and laboratory tests, in particular, to ensure that the included cases exclusively had bacterial pneumonia.

Conclusion

Our study conclusively shows that BALF-NE% enables rapid, reliable differentiation of bacterial from viral pneumonia in children, outperforming conventional diagnostic indicators. Given the nonspecific presentation and rapid progression of pediatric pneumonia, early etiological diagnosis is critical—BALF-NE% assessment provides a simple, bedside tool that meets this clinical need, facilitating timely targeted therapy and reducing antibiotic misuse.