Diagnostic agreement between bronchoalveolar lavage and transbronchial lung biopsy in patients with suspected pulmonary diseases: A single-center study

INTRODUCTION

Respiratory diseases represent a notable global health challenge, significantly contributing to morbidity and mortality rates worldwide. Traditional histopathology is considered the gold standard for diagnosing these conditions. However, cytology, specifically involving techniques such as bronchoalveolar lavage (BAL), brush cytology, and fine needle aspiration (FNA), serves as a valuable complementary diagnostic tool. Current diagnostic practices often involve obtaining patient specimens to enable accurate pathological examination, with concurrent BAL frequently performed alongside transbronchial lung biopsy (TBLB) to enhance diagnostic accuracy.^[1-4]

Evidence suggests that combining BAL with TBLB improves diagnostic precision, particularly for infiltrative peripheral lung cancers and metastatic lesions. For instance, a study by Tang et al. reported a diagnostic accuracy of 73% for infiltrative cancers when combining these techniques, rising to 100% for metastatic lung cancer.^[5] This highlights the advantages of using a dual approach to enhance diagnostic outcomes for suspected malignancies. However, limitations exist concerning non-neoplastic lesions, where the agreement between cytological findings from BAL and TBLB is often minimal. Jois et al. indicated that such limitations could potentially be addressed through the application of cytology in conjunction with lung biopsies.^[2]

While BAL is advantageous due to its minimal complication risk compared to lung biopsies, its reliability in diagnosing specific lung disorders, such as interstitial lung disease, is inconsistent and varies when compared to TBLB outcomes.^[6-9] The results of previous studies have shown that the agreement between BAL and TBLB remains notably high in cancer detection. However, there is a pressing need to better understand how these techniques can be optimized for broader lung disease pathologies. Nevertheless, under socioeconomic limitations, BAL is a quick and useful diagnostic procedure for ruling out malignancy.^[8,10] Sareen and Pandey suggested using the BAL cytology for malignancy screening before administering treatment for tuberculosis under the high prevalence of tuberculous infection region.^[8] Chang et al. even suggested that BAL could diagnose interstitial lung disease in patients with suspicious radiography and should be done early in admitted patients with malignancy, connective tissue diseases, or transplantation who had poor response to initial medications for additional treatment.^[11] These studies supported the usefulness, quickness, and cost-effectiveness of BAL diagnoses in the treatment of pulmonary diseases.

This investigation aims to assess the diagnostic agreement between BAL and TBLB in patients with suspected malignant and infectious pulmonary diseases within a single-center setting. By elucidating the relationship between these diagnostic modalities, this study aims to identify whether combining BAL with TBLB can enhance detection rates and provide a more accurate framework for diagnosing complex pulmonary conditions. Resolving this dilemma will contribute valuable insights into clinical practices and potentially improve patient management strategies in respiratory medicine.

MATERIAL AND METHODS

RESULTS

Agreement of pulmonary neoplastic diseases

Within the cohort of 104 patients suspected of having pulmonary neoplastic diseases, 56 (53.84%) were male and 48 (46.15%) female, with a mean age of 65.11 years, ranging from 39 to 87. The BAL analysis yielded positive results in 49 patients, including 34 adenocarcinomas, one squamous cell carcinoma, two non-small cell carcinomas without a specific cell type, and 12 cases that showed the presence of atypical cells (tumor type could not be specified). The TBLB led to the final diagnosis of lung cancer in 64 cases comprising 54 adenocarcinomas, six squamous cell carcinomas, three metastatic cancers (metastatic osteosarcoma, metastatic colonic adenocarcinoma, and metastatic nasopharyngeal squamous cell carcinoma), and one Kaposi sarcoma as indicated in Table 1 and Figure 1 evidencing malignancy in both BAL and TBLB. Figures 1a and b reveal pulmonary adenocarcinoma, and Figures 1c and d reveal pulmonary squamous cell carcinoma.

Table 1: Frequency of pulmonary diseases according to cytological diagnosis and histological type.

Neoplastic pulmonary diseases
Diagnosis	BAL n, (%)	TBLB n, (%)
Negative for malignancy	55 (52.88%)	40 (38.46%)
Adenocarcinoma	34 (32.69%)	54 (51.92%)
Squamous cell carcinoma	1 (0.96%)	6 (5.77%)
Atypical cells	12 (11.54%)	Metastatic tumor 3 (2.88%)
Positive for NSCLC	2 (1.92%)	Kaposi sarcoma 1 (0.96%)
Total	104 (100.00%)	104 (100.00%)
Non-neoplastic pulmonary diseases
Diagnosis	BAL n, (%)	TBLB n, (%)
Negative for malignancy or micro-organism	79 (92.94%)	62 (72.94%)
Positive for fungus	4 (4.71%)	9 (10.59%)
Positive for eosinophils	1 (1.18%)	Chronic eosinophilic pneumonia 1 (1.18%)
Positive for CMV-infected cells	1 (1.18%)	CMV pneumonitis 1 (1.18%)
	-	Granuloma 12 (14.12%)
Total	85 (100.00%)	85 (100.00%)

BAL: Bronchoalveolar lavage, TBLB: Transbronchial lung biopsy, CMV: Cytomegalovirus, NSCLC: Non-small cell lung cancer

Close

Figure 1:

Malignancy. (a and b) Positive for adenocarcinoma. (a) The bronchoalveolar lavage (BAL) revealed clusters of malignant cells with vacuolated cytoplasm (Liquid-based preparation, Papanicolaou stain, ×40 objective) Scale bar, 50 µm. (b) The transbronchial lung biopsy (TBLB) from the same case revealed tissue fragments of malignant cells with glandular formations (Hematoxylin-Eosin [H&E] stain, ×10 objective). Scale bar, 200 μm. (c and d) Positive for squamous cell carcinoma. (c) BAL revealed a group of malignant cells with coarse chromatin and a moderate amount of dense cytoplasm (Liquid-based preparation, Papanicolaou stain, ×40 objective). Scale bar, 50 μm. (d) TBLB in the same case revealed tissue fragments of malignant squamous epithelium among desmoplastic stroma (H&E stain, ×10 objective). Scale bar, 200 µm.

Export to PPT

Out of 104 patients suspected of having pulmonary neoplastic diseases, 55 were determined to be cytologically negative for malignancy by BAL, with 40 accurately identified as histologically negative for malignancy through TBLB. The overall agreement between BAL cytology and TBLB histopathological diagnosis was 70.19% with a 95% confidence interval (95% confidence interval [CI]) of 60.43– 78.76 and kappa’s statistics of 0.52 (95% CI: 0.38–0.66), indicating moderate agreement. The diagnostic accuracy of the BAL cytological examination was 79.81%. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of BAL cytology are presented in Tables 2 and 3.

Table 2: Comparative analysis of TBLB and BAL fluid cytology of pulmonary neoplasm.

Bronchoalveolar lavage	Transbronchial lung biopsy
	Negative for malignancy	Adeno carcinoma	Squamous cell carcinoma	Metastatic tumor	Kaposi sarcoma	Total	Overall agreement (95% CI)	Kappa (95% CI )
Negative for malignancy	37	11	4	2	1	55	70.19% (60.43–78.76)	0.52 (0.38–0.66)
Adenocarcinoma	0	34	0	0	0	34
Squamous cell carcinoma	0	0	1	0	0	1
Atypical cells	3	7	1	1	0	12
Positive for NSCLC	0	2	0	0	0	2
Total	40	54	6	3	1	104

BAL: Bronchoalveolar lavage, TBLB: Transbronchial lung biopsy, NSCLC: Non-small cell lung cancer, CI: Confidence interval

Table 3: Sensitivity, specificity, and diagnostic accuracy of BAL cytological examination in diagnosis of pulmonary neoplastic diseases.

Parameter	Finding
Sensitivity	93.90% (95% CI: 83.10–98.70)
Specificity	67.30% (95% CI: 53.30–79.30)
ROC area	0.82 (95% CI: 0.73–0.88)
Positive predictive value	71.90% (95% CI: 59.20–82.40)
Negative predictive value	92.50% (95% CI: 79.60–98.40)
Diagnostic accuracy	79.81%

CI: Confidence interval, ROC: Receiver operating characteristics, BAL: Bronchoalveolar lavage

Agreement on pulmonary non-neoplastic diseases

In another subset of 85 patients suspected of having nonneoplastic pulmonary diseases, 54 (63.52%) were males and 31 (36.47%) females, with a mean age of 58.14 years ranging from 17 to 89. The BAL revealed evidence of pulmonary infection in five cases (four for fungal infection and one for cytomegalovirus infection) and one case of eosinophil-rich fluid. TBLB contributed to the final diagnosis of pulmonary infection in 22 cases, including nine cases of fungal infection, one case of cytomegalovirus infection, and 12 cases of caseating granulomatous inflammation. In addition, one case was diagnosed with chronic eosinophilic pneumonia by TBLB. Table 1 and Figure 2 show details of infections in both BAL and TBLB. Figure 2a-c reveals yeasts in macrophages. Figure 2d-f reveals inflammatory cells in both Papanicolaou and H&E stains, while the acid-fast stain reveals AFB positive.

Close

Figure 2:

Infection (a-c) Histoplasmosis. (a) The bronchoalveolar lavage (BAL) revealed a macrophage containing round-shaped microorganisms (Liquid-based preparation, Papanicolaou stain, ×40 objective). (b) The transbronchial lung biopsy (TBLB) from the same case revealed granulomas (Hematoxylin-Eosin [H&E] stain, ×40 objective). (c) GMS stain revealed small narrow-based budding yeasts in macrophages (GMS stain, ×40 objective). (d, e, and f) Tuberculosis. (d) The BAL revealed inflammatory cells among fibrinous backgrounds (Liquid-based preparation, Papanicolaou stain, ×40 objective). (e) TBLB from the same case revealed numerous macrophages and epithelioid histiocytes (H&E stain, ×40 objective). (f) Numerous acid–fast bacilli (Acid–fast bacilli stain, ×40 objective). Scale bar, 50 µm.

Export to PPT

On the other hand, in 79 out of 85 cases, no evidence of infection or malignancy was diagnosed by BAL, while 62 out of 85 cases showed no evidence of infection or malignancy through histological examination. The agreement between BAL cytology and TBLB histopathology in negative results reached 77.65% (95% CI: 67.31–85.97), with a kappa coefficient of 0.29 (95% CI: 0.17–0.41), indicating fair agreement, as illustrated in Table 4.

Table 4: Comparative analysis of TBLB and BAL fluid cytology of pulmonary non-neoplastic disease.

Bronchoalveolar lavage	Transbronchial lung biopsy
	Negative for malignancy	Positive for fungus	Pulmonary eosinophilia	Positive for CMV	Granuloma	Total	Overall agreement (95% CI)	Kappa (95% CI )
Negative for malignancy or micro-organism	61	6	0	0	12	79	77.65% (67.31–85.97)	0.29 (0.17–0.41)
Positive for fungus	1	3	0	0	0	4
Positive for eosinophils	0	0	1	0	0	1
Positive for CMV	0	0	0	1	0	1
Total	62	9	1	1	12	85

BAL: Bronchoalveolar lavage, TBLB: Transbronchial lung biopsy, CMV: Cytomegalovirus, CI: Confidence interval

DISCUSSION

The incidence of pulmonary lesions has steadily increased in recent years. Clinicians and pathologists have many patients with pulmonary lesions. Patients prefer straightforward and economical examinations that result in shorter hospital stays. Preliminary screening may involve bronchial cytology (brush and wash). Pathologists and clinicians must engage in interactions and cooperation to facilitate additional treatment in pulmonary cytopathology fields.^[13] This study aims to evaluate the diagnostic agreement between BAL cytology and TBLB histopathologic analysis in patients with suspected pulmonary conditions. The results demonstrate a substantial level of agreement (70.19%) with a kappa statistic of 0.52 for neoplastic diseases. These findings indicate that combining BAL and TBLB can effectively diagnose lung malignancies and support their use as complementary diagnostic tools in clinical practice.

Several factors contribute to the agreement between BAL cytology and TBLB histology for pulmonary neoplastic diseases. Dionísio reveals that the sensitivity of TBLB for diagnosing peripheral pulmonary neoplastic lesions fluctuates based on the number of biopsy samples extracted and the size of the lesion.^[14] Moreover, the distribution of the lesion, whether focal or diffuse, combined with the small size of the acquired samples, the potential for confounding factors arising from crush artifacts, and the challenge of penetrating beyond the peribronchial sheath are all significant considerations.^[15]

In comparing our findings with previous studies, it is noteworthy that the sensitivity and PPV of BAL in our cohort were higher than those reported by Salama et al., who found a sensitivity of only 40.3% and a specificity of 51.7%.^[16] While our study shows improved sensitivity and PPV, it also reflects lower specificity and NPV. Moreover, a study by Ahmed and Ahmed reported higher sensitivity of BAL cytology (93.44%), specificity (100%), and NPV (75%), with 94.5% diagnostic efficacy compared with transbronchial needle aspiration.^[10]

This highlights the diverse nature of existing research and suggests a need for careful interpretation of diagnostic metrics in different clinical contexts.

The moderate agreement (77.65%) observed for nonneoplastic diseases, with a kappa statistic of 0.29, underscores the complexities involved in diagnosing pulmonary conditions that are not malignancies. While this suggests some consistency between BAL and TBLB, the relatively low kappa value indicates fair to poor agreement, suggesting that these methods should be interpreted cautiously in nonneoplastic scenarios. Factors influencing low agreement levels could include the distribution, size, and complexity of the non-neoplastic pulmonary diseases evaluated, as well as the inherent limitations of both diagnostic techniques.

Various factors, such as the distribution of lesions, the size and type of non-neoplastic pulmonary diseases being assessed, and the specific methodologies used in BAL and TBLB, may influence the level of agreement observed. Jois’s study examined the correlation between TBLB and lung cytology in respiratory specimens. The highest level of concordance for malignancy was observed with imprint cytology, achieving a rate of 77.78%, followed by BAL at 40.91% and brush cytology at 40.00%. Overall, the degree of agreement between cytological evaluations and biopsy results was significantly low for non-neoplastic lesions. Notably, when BAL and imprint smears were utilized in tandem, the concordance rate improved to 42.50% compared to other methodological combinations. The authors advocate the application of a multimodal approach incorporating various cytological techniques in cases where malignancy is suspected, as this strategy is more efficacious than relying on a single diagnostic modality. They recommend the systematic inclusion of imprint smears across all cases; however, it remains clear that biopsy continues to be the gold standard for the diagnosis of non-neoplastic lung diseases.^[2] In addition, studies from Davidson et al. and Tomassetti et al. concluded that BAL with cellular analysis was found to be a useful diagnostic adjunct in interstitial lung diseases.^[1,4]

The moderate agreement between BAL cytology and TBLB histopathology has important implications for clinical decision-making. While BAL demonstrates a high diagnostic accuracy, its limitations in sensitivity suggest that it should not be solely relied upon for malignancy detection, especially in patients with high suspicion. This necessitates a multifaceted diagnostic approach, incorporating both BAL and TBLB for enhanced diagnostic precision. In addition, understanding the potential for false negatives in BAL emphasizes the importance of histological confirmation through TBLB as the gold standard.

One of the primary strengths of our study lies in its design, involving a comprehensive retrospective analysis of bronchoscopic procedures conducted at a single institution. The rigorous methodology adopted, including well-defined inclusion criteria and thorough data collection, contributes to the reliability of our findings. In addition, the technical proficiency of the pulmonologists performing the procedures likely enhances the accuracy of the results, thereby improving the interpretation of diagnostic outcomes.

Despite these strengths, certain limitations must be acknowledged. The specificity of BAL for neoplastic diseases was found to be relatively low (67.3%), potentially influenced by factors such as the location of peripheral tumors, which might yield fewer detectable malignant cells, and the presence of degenerative cells that could confound results. Such limitations are consistent with existing literature, highlighting the variability in diagnostic performance due to sampling techniques, the experience of the bronchoscopist, and the distribution and histological characteristics of the lesions involved.^[10,14,16] In addition, the higher adverse event rate noted in this study may reflect the severity of the pulmonary conditions in the patient cohort under investigation. While our agreement percentage of 70.19% indicates moderate concordance, it is critical to acknowledge that kappa statistics also account for chance agreements, which can obfuscate true diagnostic reliability. Moreover, the single-center design may restrict the generalizability of the results, as data derived from one institution may not accurately represent the diversity of patient demographics, disease prevalence, or treatment practices observed in multicenter studies. The sample size of 189 patients, although adequate for initial analyses, may limit the robustness of conclusions, particularly concerning specific pulmonary disease subgroups, as a larger and more diverse cohort would yield a more accurate representation of diagnostic outcomes.

Future research should address these limitations through multicenter trials with larger sample sizes and standardized methodologies while also exploring the potential impact of technological advancements in cytology and histopathology, such as the implementation of artificial intelligence in image analysis and enhanced imaging techniques, which could improve diagnostic accuracy and agreement between BAL and TBLB. Integrating these innovations may lead to more reliable diagnoses and improved patient outcomes.

SUMMARY

This research highlights the value of BAL cytology in diagnosing pulmonary carcinoma and pulmonary infections. In the context of diagnosing lung cancer, a TBLB procedure is essential for accurate tissue examination. Despite BAL’s potentially low yield, its inclusion in the procedure can be beneficial, potentially saving time. Furthermore, the examination of BAL fluid may enable the diagnosis of pulmonary diseases in patients with inaccessible peripheral pulmonary lesions. Combining TBLB with BAL has the potential to further enhance the positive diagnostic rate.

AVAILABILITY OF DATA AND MATERIALS

The datasets used and/or analyzed during the present study are available from the corresponding author upon reasonable request.

ABBREVIATIONS

AFB: Acid–fast bacilli

BAL: Bronchoalveolar lavage

FN: False negative

FNA: Fine needle aspiration

FP: False positive

GMS: Gomori-methenamine silver

H&E: Hematoxylin-eosin

NPV: Negative predictive value

PPV: Positive predictive value

TBLB: Transbronchial lung biopsy

TBNA: Transbronchial needle aspiration

TN: True negative

TP: True positive

AUTHOR CONTRIBUTIONS

KW: Concept, design, literature search, clinical studies, data acquisition, data analysis, statistical analysis, manuscript preparation, manuscript editing, manuscript review, and funding; SK and NL: Clinical studies and data acquisition; PP: design, data analysis, and statistical analysis; CN, TY, PS, PV, and SC: Literature search, data acquisition, data analysis; WD: Literature search, clinical studies, data acquisition, data analysis, statistical analysis, manuscript preparation, manuscript editing, and manuscript review; SCH: Concept, design, literature search, clinical studies, data acquisition, data analysis, statistical analysis, manuscript preparation, manuscript editing, and manuscript review. All authors have provided their final approval for the version intended for publication. Each has participated adequately in the work to assume public responsibility for relevant parts of the content and has agreed to be accountable for all aspects of the work, ensuring that any issues regarding its accuracy or integrity are addressed. All authors meet the authorship status of ICMJE.