Value of bronchoscopy after computed tomography in the diagnostic work-up of haemoptysis

ABSTRACT

Introduction

Haemoptysis, a common symptom with a variety of aetiologies, requires a comprehensive diagnostic approach. Computed tomography of the thorax (CT) and bronchoscopy are crucial in this process. This study assesses the value of bronchoscopy in patients with haemoptysis and CT with no findings suspicious of malignancy.

Methods and materials

This study retrospectively assessed patients with haemoptysis referred between January 2019 and December 2022 to the Department of Respiratory Medicine, Odense University Hospital. All patients who underwent CT and subsequently bronchoscopy due to haemoptysis were included in the study. Patients were excluded if CT or bronchoscopy was not performed, as well as patients suspected of malignancy on CT. Data, including patient characteristics, medical history, haemoptysis progression, CT, bronchoscopy results, and clinical assessments, were collected with a six-month follow-up.

Results

A total of 469 patients were assessed for eligibility, of whom 306 were excluded, resulting in 163 eligible patients. There were 95 males (58.28%), and the mean age was 57.8 years (SD 14.48). Mild haemoptysis (92.02%) and longer than one week (63.80%) was the most common presentation. CT revealed abnormal findings in 77 (47.24%) patients and for 117 (71.78%) patients followed a bronchoscopy with no abnormal observations. No malignant findings were detected by bronchoscopy.

Conclusion

The findings of this study revealed that bronchoscopy did not identify malignant aetiology beyond the scope of CT. These findings combined with current evidence suggest minimal benefit of bronchoscopy subsequent to CT in the diagnostic evaluation of thoracic malignancy in these patients. Take Home Message The diagnostic approach for patients with haemoptysis typically includes computed tomography (CT) and bronchoscopy to exclude pulmonary malignancy. This study supports current evidence of minimal value of bronchoscopy subsequent to CT in the diagnostic evaluation of thoracic malignancy in these patients.

Introduction

Haemoptysis

Haemoptysis, the expectoration of blood from the respiratory tract, poses a diagnostic challenge due to its diverse aetiologies and potential clinical significance [1, 2]. The severity of the aetiology leading to haemoptysis may vary from cancer to pulmonary infection and further diagnostic tests, treatment and outcome consist of a wide range of scenarios [1-6].

The quantity of haemoptysis is although challenging, a key element of the diagnostic and therapeutic approach as mild haemoptysis is evaluated through outpatient clinic, while severe haemoptysis is often life-threatening and requires admission to intensive care [7]. Categorization of haemoptysis is widely discussed and no international consensus is established, but studies have shown that 95% of the cases with haemoptysis are mild, while the majority of the cases are remitting without medical intervention [8–10].

Essentially mild and moderate amounts of haemoptysis must be distinguished from severe amounts, as severe haemoptysis requires imminent evaluation and treatment to prevent high mortality [11]. The diagnostic work-up of mild haemoptysis centres on exclusion of severe aetiologies such as cancer. As haemoptysis is a symptom with high risk of underlying lung cancer, all patients regardless of age and history of smoking should be referred for computed tomography of the thorax (CT), especially patients of age above 40 [12–15]. Patients with suspected malignant CT findings (such as solid lung lesion, lung nodules with high risk of malignancy, carcinomatosis of the lungs, or metastasis from other malignancies) are referred directly for diagnostic work-up for lung cancer [13,16,17]. CT findings necessitate a thorough evaluation to ensure correct diagnostic assessment. Observed solid lung lesions > 30 mm are usually considered malignant until proven otherwise or if obvious benign genesis is presented (e.g. atelectasis or infection).

Patients with haemoptysis without suspicious malignant findings on CT are referred to a pulmonary department for further assessment. The assessment includes thorough medical history and clinical examination in order to determine the amount and time span of the bleeding, to exclude oropharyngeal or gastrointestinal source or to identify other obvious reasons for haemoptysis. Since endobronchial tumors are not always detected by CT, bronchoscopic examination is used to refute suspicion of malignancy in patients with undetermined cause for haemoptysis [7,18,19].

Studies have indicated that bronchoscopy may be deemed unnecessary when CT reveals no indication of malignancy [20–22]. This procedure often provides discomfort and a psychological burden to the patient but is also expensive and time-consuming. Thus, scrupulous patient selection for bronchoscopic examination would minimise patient discomfort and risk in the diagnostic work-up of haemoptysis, and additionally reduce resource use at an organizational level. However, the literature is divergent on the matter, and current national and international guidelines still suggest a diagnostic pathway including both CT and bronchoscopy for these patients.

Aim and research objectives

The primary aim of this study is to evaluate the value of bronchoscopy for diagnosing malignancy in patients with haemoptysis and no findings suspicious of malignancy on a CT. The secondary aim of this study is to describe clinical, radiological and bronchoscopy findings for all patients assessed for haemoptysis.

Methods and materials

Setting and study population

The study is a retrospective, single-center descriptive study. Consecutive patients referred to the Department of Respiratory Medicine, Odense University Hospital (OUH), between January 2019 and December 2022 with haemoptysis were assessed for eligibility. Patient records from the electronic patient journals (EPJ) combined with available imaging were used to assess patient eligibility and to obtain data.

Inclusion and exclusion criteria

Patient eligibility was based on the following inclusion and exclusion criteria:

Inclusion: Patients referred to the Department of Respiratory Medicine, OUH, due to haemoptysis.

Exclusion:

1. No CT performed

2. No bronchoscopy performed

3. Bronchoscopy performed before CT

4. Bronchoscopy performed due to CT with findings suspicious of malignancy

5. Missing EPJ information.

Assessment of malignancy by CT and bronchoscopy protocol

The protocol for imaging consists of a contrast enhanced CT of thorax and abdomen with Omnipaque 350 mg/m. An axial 0.63 mm and 1.25 mm reconstruction was used for soft tissue and lung, respectively. A 3 mm reconstruction was used for coronal soft tissue and sagittal plane.

Exclusion criteria 4) is based on national and international guidelines [16,17,23–26]. Stratification of patients regarding nodules was done accordingly to the following:

CT suspicious of malignancy:

1. Solid nodule > 8 mm with a risk of malignancy ≥10% according to Brock model [27].

2. Solid nodules with Volume Doubling Time < 400 days assessed in combination with Brock and Herders model.

3. Sub solid nodule with progression such as change in morphology or increasing component of ground-glass or solid part.

Patient CT with findings of pulmonary nodule, but none of the abovementioned characteristics (I-III) were classified as not suspicious of malignancy and were either booked for follow-up CT or terminated from further diagnostic evaluation. Thereby being eligible for inclusion if no other exclusion criteria was observed.

If any lesions suspicions of malignancy were reported by the radiologist, a FDG positron emission tomography CT (PET-CT) was subsequently performed before further assessment by bronchoscopy.

Bronchoscopy was performed with moderate sedation of the patient and with a flexible bronchoscope. A systematic view of the bronchial tree to a subsegmental level was performed. Depending on the macroscopic findings, physician was able to collect biopsy, BL and/or BAL.

Data extraction

Data extraction was performed by two investigators. Variables were registered in a SharePoint Excel spreadsheet. Patients eligible for inclusion in the study were registered, and patient demographics including age, gender, smoking status (active, former or never), pack-years, comorbidities and prescription of anticoagulation and antibiotics were extracted from EPJ.

Details on the course of haemoptysis were collected from EPJ. Haemoptysis was characterized according to the following variables: mild ( < 31 ml/day), moderate (31–411 ml/day) and severe (≥411 ml/day) according to national guidelines, duration ( < 1 week or ≥1 week), active haemoptysis ( < 48 h), relapse, as well as symptoms of an oropharyngeal source of bleeding [28].

Data on CT findings were collected from radiographic descriptions, including diffuse unilateral or bilateral parenchymal involvement, nodules, nodule morphology and size (≤8 mm or > 8 mm).

Bronchoscopic information regarding macroscopic findings (tumour, source of bleeding, malignancy and other), possible bronchial lavage (BL) and bronchioalveolar lavage (BAL), with microbiological and cytological results, were collected from the patient journals.

A six-month follow-up of all patients´ medical records was performed to obtain information regarding subsequent malignant diagnosis.

Statistics

Descriptive data was analysed using Excel descriptive statistics program, and further analysis was conducted with the use of R v. 4.2.1. Normality of continuous variables was tested with the Shapiro-Wilk test. Normally distributed variables were expressed as mean and standard deviation, non-normally distributed variables as median and interquartile range. T-test was used to compare normally distributed, unpaired data, while Wilcoxon rank-sum test for not-normally distributed, unpaired sample. A two-sided Fisher’s exact test was performed on categorical variables. A p-value less than 0.05 was considered statistically significant.

Results

Study population and haemoptysis symptoms

A total of 469 patients were screened for eligibility. Exclusion of patients before CT (n = 129), after CT (n = 154), and because of suspected malignancy (n = 22), resulted in 163 eligible patients (figure 1). All patients with haemoptysis but no radiological suspicion of malignancy were offered a bronchoscopy after the CT according to the national Danish guidelines, with the exception of 25 patients, where extrapulmonary source of bleeding was identified and 32 patients that were not eligible for bronchoscopy due to eastern cooperative oncology group performance status (PS) of 3–4; however, 97 patients did not wish a bronchoscopy and chose not to proceed with the examination. Baseline characteristics of the included patients are given in Table 1. The mean age of patients was 57.8 years (SD 14.48), with 95 (58.28%) of them being men. Comorbidities were described in 21 (12.88%) of the patients, which primarily consisted of chronic obstructive pulmonary disease COPD (n = 17, 10.43%), while 20 (12.69%) patients had been previously diagnosed with other malignancy. Patients’ PS was evaluated to be 0 and 1 in the majority of the cases, respectively, 105 (64.42%) and 33 (20.25%) patients. History of smoking, either active or former, was recorded in 116 (71.1%) patients with median of 10 pack-years. A small proportion of the population was on anticoagulant (n = 17, 10.43%) or antiplatelet (n = 21, 12.88%) medicine.

Figure 1.

Flowchart of patients inclusion.

Flowchart of inclusion and exclusion. Created with Biorender.

Table 1.

Baseline patient characteristics.

Age (mean)	57.8 years (SD 14.48)
Sex, male (%)	95 male (58.3%)
Comorbidities
COPD	n = 17 (10.43%)
ILD	n = 1 (0.61%)
Bronchiectasis	n = 3 (1.84%)
WHO-PS
PS 0	n = 105 (64.42%)
PS 1	n = 33 (20.25%)
PS 2	n = 6 (3.68%)
PS 3	n = 2 (1.23%)
PS 4	n = 1 (0.61%)
Smoking status
Active	n = 45 (27.60%)
Former	n = 71 (43.56%)
Never	n = 47 (28.83%)
Smoking pack-years	10 years (IQR 0–30)
Medicine
Anticoagulation	n = 17 (10.43%)
Antiplatelet	n = 21 (12.88%)
Former conditions
Cancer	n = 20 (12.69%)
Tuberculosis	n = 6 (3.68%)
Mucosal infection	n = 2 (1.23%)
Active haemoptysis	n = 64 (39.26%)
Duration
< 1 Week	n = 58 (35.58%)
≥1 Week	n = 104 (63.80%)
Quantification
Mild	n = 150 (92.02%)
Moderate	n = 9 (5.52%)
Severe	n = 3 (1.84%)
Oral or nasal bleeding	n = 18 (11.04%)
Relapse of haemoptysis	n = 41 (25.15%)
Associated pulmonary infection	n = 22 (13.49%)
Treated with antibiotics	n = 14 (8.59%)

Baseline characteristics indicated by number (n) of patients and percentage (%) of total sample size.

IQR = Interquartile Range. SD = Standard Deviation

Patients dominantly described ≥1 week duration (n = 104, 63.80%), and mild level of haemoptysis (n = 150, 92%). Pulmonary infection was suspected in 22 (13.49%) patients on the basis of primary anamnestic information (Table 1).

CT findings

CT revealed that nearly half (n = 77, 47%), of the patients had abnormal pulmonary findings. Diffuse pulmonary parenchymal abnormalities were described for 56 (34.36%) patients, of which bilateral involvement was present in 21 (37.5%). Pulmonary nodules were identified in 30 (18.40%) patients and small nodules (≤8 mm) were observed in 17 (56.67%) of them (Table 2).

Table 2.

CT findings.

CT with abnormal findings	n = 77 (47.24%)
Diffuse pulmonary parenchymal abnormality	n = 56 (34.36%)
Bilateral involvement	n = 21 (37.50%)
Nodule	n = 30 (18.40%)
Nodule size
≤8 mm	n = 17 (56.67%)
> 8 mm	n = 13 (43.33%)
Nodule morphology
Solid	n = 12 (40.00%)
Part solid	n = 13 (43.33%)
Both solid and part solid	n = 1 (3.33%)
Cystic	n = 1 (3.33%

Results (n) indicating amount of patients and percentage of total sample size or subcategory. Abbreviations: CT = Computed Tomography; mm = millimetres.

Bronchoscopy

Subsequent bronchoscopy revealed no abnormal findings in 117 (71.8%) of the cases, while no tumours were identified (Table 3). The source of bleeding was identified in two (1.2%) patients. Observation of other pathology was registered in 44 (27%) patients with vulnerable mucosal membrane being the most common finding (n = 25, 15.3%) (Table 3 and Supplementary Table S2).

Table 3.

Bronchoscopy findings.

	Abnormal CT	Normal CT
Observations during bronchoscopy
Normal bronchoscopy	n = 53(45.3%)	64 (54.7%)
Source of bleeding identified	n = 2 (100%)	n = 0 (0%)
Other	n = 24 (54.5%)	n = 20 (45.6%)
Vulnerable mucous membranes	n = 8 (32.0%)	17 (78.0%)
Samples
BL performed	n = 40 (27.2%)	n = 107 (72.8%)
BAL performed	n = 26 (66.715.95%)	n = 13 (33.3%)
Patients with positive microbiological findings	n = 10 (28.6%)	n = 25 (71.4%)
TBB	n = 1 (12.5%)	n = 7 (87.7%)
Unspecific	n = 1 (20.0%)	n = 4 (80.0%)
Interstitial findings	n = 0 (0%)	n = 2 (100%)
Eosinophilia	n = 0 (0%)	n = 1 (100%)

Abbreviations: BL = Bronchial Lavage; BAL = Bronchoalveolar Lavage; TBB = Transbronchial Lung Biopsy.

Results of bronchoscopy lead to a change in treatment in 17 (10.4%) patients (Supplementary Table 2). Of 148 patients investigated microbiologically during bronchoscopy, 35 (23.7%) had a positive result of which initiated or changed prescription of antibiotics to 10 (28.6%) patients (Supplementary Table S3–S4). A post hoc analysis of patients with positive and negative microbiological findings from bronchoscopic samples revealed no significant differences in age, PS, former conditions, or medication. A significant difference was observed in smoking status, revealing a higher level of positive microbiological findings in patients with smoking history (Table 4).

Table 4.

Comparison of patient characteristics in regard to microbiological results.

	Positive microbiological result	Negative microbiological result	P-value
N	35	113
Age (mean)	59	57.4	p = 0.564
Gender
Male	48.6% (n = 17)	61.1% (n = 69)	p = 0.571
Female	51.4% (n = 18)	38.9% (n = 58)
ECOG-PS
0	68.6% (n = 24)	65.5% (n = 74)	p = 0.251
1	22.9% (n = 8)	19.5% (n = 22)
2	0% (n = 0)	2.9% (n = 3)
3	2.9% (n = 1)	0.9% (n = 1)
4	2.9% (n = 1)	0% (n = 0)
Smoking status
Active	45.7% (n = 16)	22.1% (n = 25)	p = 0.024
Former	31.4% (n = 11)	46.0% (n = 52)
Never	22.9% (n = 8)	31.9% (n = 36)
Package years
Median or mean*	22.03*	22	p = 0.01
Anamnestic conditions/Comorbidities
Cancer	14.3% (n = 5)	3.5% (n = 4)	p = 0.990
Tuberculosis	5.7% (n = 2)	1.8% (n = 2)
Mucosal infection	0% (n = 0)	0% (n = 0)
COPD	8.6% (n = 3)	1.8% (n = 2)
ILD	0% (n = 0)	0% (n = 0)
Bronchiectasis	2.9% (n = 1)	0.9% (n = 1)
Medicine
Anticoagulation	5.7% (n = 2)	1.8% (n = 2)	p = 0.997
Antiplatelet	11.4% (n = 4)	3.5% (n = 4)
Antibiotics	11.4% (n = 4)	3.5% (n = 4)

Abbreviation: ECOG-PS = European Cooperative Oncology Group Performance Status; COPD = Chronic Obstructive Pulmonary Disease; ILD = Interstitial Lung Disease.

Aetiology of haemoptysis

In correspondence to the primary endpoint, bronchoscopy did not show any malignancy beyond the scope of the CT. Six-month follow-up revealed one patient with T1bN0M0 pulmonary adenocarcinoma, while three excluded patients had pulmonary malignancy (Figure 2 and Supplementary Table S5). The one patient included where pulmonary malignancy was revealed within the follow-up period was being regularly followed with thoracic CT every six months, due to a 14 × 19 mm part solid nodule. Six months after evaluation for haemoptysis, a renewed CT showed progression of the same nodule, now estimated to 24 × 16 mm with increasing solidity.

Figure 2.

Flowchart malignant vs. benign diagnosis with six month follow-up.

Follow-up of patients included and excluded. Created with Biorender.

The majority of the included patients (n = 122, 74.9%) had cryptogenic haemoptysis, while 23 (14.7%) patients were diagnosed with pulmonary infection. Other aetiologies were present in 18 (10.4%) patients (Supplementary Table S3). Of the patients included, 11 (6.6%) were referred to the department again due to haemoptysis, with a mean of 22.8 months (SD 9.41) time to re-referral. There was no statistically significant difference in the risk factors for recurrence – including age, gender, performance status, smoking status, pack years, comorbidities, and medication – between patients who were re-referred and those who were not.

Discussion

The main observation of this study is that no malignancy was found by bronchoscopy following CT with no sign of cancer. Thus, CT provided no further diagnostic value in the evaluation of pulmonary malignancy in patients with haemoptysis.

In alignment with recent pertinent literature, several studies have investigated the role of bronchoscopy in the diagnostic work-up of haemoptysis.

Four studies by Petersen et al., Nielsen et al., Bønløkke et al., Tsoumakidou et al. and Arooj et al. have all identified that no additional malignancy was diagnosed due to bronchoscopy subsequent to CT [21,29–31]. Studies from Lee et al. and Thirumaran et al. both report one missed malignant diagnosis by CT, respectively, 1/228 and 1/270, which was diagnosed due to bronchoscopy [32,33]. Older studies by McGuiness et al. describe 6 cases of lung cancer identified by bronchoscopy in 57 haemoptysis patients, while Set et al. revealed that 37 of 91 patients with haemoptysis presented with malignant diagnosis. The higher incidence of lung malignancy reported in these papers are the result of significant difference in study design compared to this study, as patients presented with haemoptysis regardless of CT being indicative of malignancy were included. A meta-analysis by O´Mahoney et al. aimed to address whether bronchoscopy offers additional value to rule-out lung cancer in patients with haemoptysis and a negative CT. After including 14 studies and a total of 2,960 patients, they report only three cases with false negative CT, of which one had a positive bronchoscopy [34]. These findings are supported by a more recent systematic review and meta-analysis by Mohammad et al., which strongly support the minimal benefit of subsequent bronchoscopy [35].

In the present study, bronchoscopy lead to non-malignant findings associated with haemoptysis in 27% of the patients. Bronchoscopy was valuable to correctly diagnose benign aetiology and led to a change of treatment due to prescription, change or stop of antibiotics in 11 patients. Initiating treatment with antibiotics was described almost exclusively on the basis of microbiological findings (8/9) as only one patient was described to have symptoms. This could be problematic as treatment to asymptomatic patient could be avoided; however, information on symptoms was collected retrospectively and may be underreported. Positive microbiological findings from bronchoscopy were significantly associated with current or previous smoking status and increased package years. To our knowledge, this has not been reported previously. Petersen et al. report 24.9% positive findings for the patients that underwent microbiological sampling but do not further comment on the patient characteristics.

Patient demographics presented in previous studies are similar to our findings, where a median age between 44 and 60.5 years and an overrepresentation of male patients is observed [20–22,30,33]. This finding of mild haemoptysis is in concordance with the literature [22,30,33]. The classification of mild, moderate and severe haemoptysis in this study is based on national guidelines and could therefore differ from future studies, as no international agreement on classification is published [28]. Notably, if a different classification on the basis of quantitative amount of haemoptysis, such as Quiqley et al., amount of patients with severe haemoptysis could prove higher in this study [36].

Approximately half of the study population exhibited CT abnormalities, none of which were indicative of malignancy. While abnormal CT findings are commonly observed in haemoptysis patients, the reported incidence differs according to study methodology [29,30]. Lee et al. have observed nodules in 33% of the CT compared to 18.40% in this study, which may be explained by the inclusion of granulomas and a larger sample size [37]. Nodule-assessment by Brock’s model is therefore an important clinical tool to assess these patients. The clinical work-up of our study population is also reflected in CT findings, as malignant findings were excluded due to the endpoint of the original study design. Furthermore, this exclusion could influence the results, as patients would undergo FDG PET-CT and biopsy and not referred to ordinary bronchoscopy.

In this study, the source of bleeding was identified at only 1.2% of the cases. Data from Revel et al. showed that 64% of the bronchoscopies performed visualized side and segment or lobe of bleeding, which is explainable due to patients included being admitted to the intensive department [38]. The main part of this study´s patients were referred to the outpatient clinic with a mild haemoptysis and time of diagnostic work-up was presumably much longer than Revel et al., ultimately resulting in higher incidence of self-limiting haemoptysis.

Cryptogenic was predominantly described as the cause of haemoptysis in our study. Current literature is diverse in reporting of aetiology in patients with haemoptysis. Studies similar to ours describe cryptogenic aetiology to be most common followed by infectious disease [20,22]. Studies who do not risk stratify patients based on CT find a higher incidence of malignancy, although not superior to infectious disease [21,33,38].

One patient included in this study was diagnosed with lung cancer within the follow-up period. The patient was referred to control CT after 6 months due to 9 mm solid nodule. The control CT showed growth of the nodule and further diagnostic investigation revealed pulmonary adenocarcinoma. However, it is unlikely that the 9 mm nodule was the source of haemoptysis. Both this study´s findings and available data emphasize the clinical importance of CT as first-line diagnostic tool for evaluation of these patients [34].

Of the patients included in this study, only 6.8% were re-referred due to recurrence of haemoptysis. In comparison, other studies report a higher incidence of recurrence, ranging from 8.7% to 35.1%. Regarding the time to recurrence, the current literature indicates a shorter mean time to recurrence (5–15 months) compared to the 22.8 months observed in this study [29,32,39]. The dissimilarity observed in this study could be due to a restricted sample size and a predominance of patients presenting with mild haemoptysis. Additionally, the analysis of risk factors associated with haemoptysis recurrence did not identify any significant factors, consistent with the findings of Mondoni et al. [40].

The present study has several strengths, including a well-defined and fairly large patient cohort with several high-quality, comprehensive information, combining patient and haemoptysis characteristics with detailed CT, bronchoscopic descriptions and follow-up data. However, it is important to acknowledge certain limitations, mainly its retrospective and single-center design with lack of standardization in procedures. Potential local variations in diagnostic approaches might also impact the generalizability of our findings.

The optimal study to assert the research objective would be a randomized controlled trial of patients with haemoptysis and CT scans with no suspicion of malignancy or pulmonary disease. Randomization would be split into two groups, with or without subsequent broncschopy, with a five-year follow-up of malignancy and other pulmonary non-malignant findings.

Conclusion

Our study results suggest that the use of bronchoscopy in cases with haemoptysis and CT without sign of lung cancer does not provide additional findings of pulmonary malignancy. A substantial proportion of the patients did benefit from bronchoscopy with non-malignant findings. Although no larger prospective studies have been conducted on this matter, current evidence, including two meta-analysis, reveals marginal value of bronchoscopy subsequent to CT in the exclusion of malignancy in these patients, and therefore suggestible that it is withheld from routine diagnostic work-up of patients not suspected of malignancy on CT.

Disclosure statement

No potential conflict of interest was reported by the author(s).

ADJ has nothing to declare.

IA has nothing to declare.

CBL has nothing to declare.

AA has nothing to declare

VP has nothing to declare.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/20018525.2025.2573588