Introduction
Pneumothorax is a critical complication that must always be considered during lung biopsy procedures. The incidence of pneumothorax associated with computed tomography (CT)-guided percutaneous needle biopsy (PCNB) is approximately 20%, whereas, for transbronchial lung biopsy (TBLB), it ranges from 1% to 8% [1-4]. In most instances, pneumothorax resolves spontaneously and can be managed conservatively with supplemental oxygen and immobilization in a position with biopsy side down [5]. However, progression of pneumothorax requires the insertion of a chest tube in about 6.6% of patients who develop post-biopsy pneumothorax [6,7]. Notably, a delayed diagnosis of pneumothorax increases the risk of needing interventional treatment, with rates of chest-tube insertion ranging from 7.1% to 29.6% [8]. Thus, prompt diagnosis is crucial.
Pneumothorax is typically diagnosed using CT or chest radiography (CXR). However, these methods pose a risk of radiation exposure, and CXR exhibits relatively low sensitivity, particularly in the supine position during the early phase. In one meta-analysis, the sensitivity and specificity of CXR performed in the supine position for detecting pneumothorax were 50.2% and 99.4%, respectively [9]. Additionally, the necessity of transporting patients to radiology suites creates logistical challenges and limits immobilization, possibly worsening the condition. Since Wernecke et al. [10] first reported using ultrasonography for diagnosing pneumothorax in 1987, lung ultrasonography (LUS) has been included in diagnostic protocols such as the Extended Focused Assessment with Sonography in Trauma and Point-of-Care Ultrasound [10-12]. Prior studies consistently demonstrate the high sensitivity of ultrasonography for detecting pneumothorax in critically ill patients, with reported sensitivity and specificity of 86%-98% and 97%-100%, respectively [13]. These findings support the reliability of LUS as a diagnostic tool in acute-care settings. Consequently, recent guidelines recommend utilizing ultrasonography over anteroposterior CXR for the diagnosis of pneumothorax in critical care settings [14].
Nevertheless, employing ultrasonography to detect pneumothorax in general ward patients remains challenging, primarily due to issues related to the size and availability of sonographic equipment in these settings. The development of handheld portable ultrasonography devices has notably enhanced the utility of this technology, allowing bedside assessments, particularly beneficial immediately post-biopsy.
Therefore, this prospective study aimed to compare the diagnostic sensitivities of handheld ultrasonography and CXR in detecting early pneumothorax, which occurs within a few hours post-biopsy, in patients undergoing lung biopsy for pulmonary lesion diagnosis.
Materials and Methods
1. Study design and patients
This prospective observational study was conducted at Seoul National University Bundang Hospital from May to October 2021, adhering to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. Patients aged >18 years undergoing PCNB, TBLB, or electromagnetic navigation bronchoscopy (ENB) for pulmonary lesions were included. Exclusion criteria were patients who declined participation, altered their diagnostic plan during hospitalization, had a history of pleurodesis, or exhibited a negative lung-sliding sign on initial LUS. Post-biopsy, patients were positioned with the biopsy side downward to reduce pneumothorax risk and monitored in the general ward for at least 24 hours. LUS and CXR were performed at 3 and 24 hours post-biopsy to detect pneumothorax (Figure 1). A diagnosis of pneumothorax was made if detected on LUS or CXR at 3 or 24 hours post-biopsy.
2. LUS and CXR
All ultrasonographic examinations were conducted using a handheld portable ultrasonography device, equipped with a linear array probe operating at a frequency of 3.3 to 8.0 MHz (Vscan Extend with Dual Probe, General Electric, Northville, MI, USA). The two-dimensional mode and standard lung preset were utilized. An internal medicine specialist with more than 1 year of LUS experience, blinded to clinical and CXR findings to prevent bias, performed all LUS examinations. Examinations took place in the supine position to assess the anterior, lateral, and posterior thoracic regions, comparing bilateral ultrasonography images. If lung sliding, lung pulse, or B-lines were absent, the transducer was rotated counter-clockwise in the intercostal space and moved laterally to determine if a lung point could be identified. Pneumothorax was determined by the absence of lung sliding (or lung pulse) and B-lines, and the presence of a ‘lung point’ [15]. Pneumothorax confirmation was conducted through standard posteroanterior CXR in a standing position, and assessed by a radiologist, who was also blinded to LUS results and clinical or procedural details.
3. Statistical analysis
Continuous and categorical data are presented as mean±standard error or number (percentage). Continuous variables were compared using Student’s t-test, and categorical variables were analyzed using the chi-square test or Fisher’s exact test. We calculated the sensitivities, specificities, positive predictive values, and negative predictive values (NPVs) for both diagnostic tools. Additionally, the accuracy of LUS and CXR in detecting pneumothorax was quantified using the area under the receiver operating characteristic curve. We compared the results of the two procedures using De Long’s test. Statistical analyses were conducted using SPSS version 23.0 for Windows (IBM Co., Armonk, NY, USA) and MedCalc version 22.014 for Windows (MedCalc, Ostend, Belgium), with statistical significance set at p<0.05.
4. Ethical considerations
This study was conducted in accordance with the principles of the Declaration of Helsinki and received approval from the Institutional Review Board of Seoul National University Bundang Hospital (approval No. B-2005-612-305). Participants were recruited from May 1, 2021, to October 31, 2021. Informed consent was obtained from all participants. Written consent was secured following a detailed explanation of the study procedures, with each form signed and dated by the participants.
Results
During the study period, a total of 118 patients underwent invasive biopsy for lung lesions. After applying predefined exclusion criteria, 32 patients were excluded, leaving 86 patients in the final analysis. Postoperative pneumothorax was confirmed on LUS or CXR in 23 patients (Figure 2). There were no statistically significant differences in sex, age, lung function, or duration of hospitalization between the pneumothorax and non-pneumothorax groups (Table 1).
In the stratified analysis based on diagnostic modalities, pneumothorax occurred exclusively in patients who underwent PCNB. There were no statistically significant differences in the incidence of pneumothorax by location of the lung lesion, presence of emphysema, or pleural effusion (p=0.39, p=0.93, and p=0.26, respectively). Although the distance between the lesion and the pleura tended to be greater in the pneumothorax group than in the non-pneumothorax group, the difference was not statistically significant (11.43±2.79 vs. 6.54±1.37, p=0.09) (Table 2). None of the patients who developed pneumothorax after biopsy required a chest-tube insertion. Differential detection of pneumothorax was noted at two post-biopsy checkpoints in the 23 patients diagnosed with pneumothorax using sequential LUS and CXR. At 3 hours post-biopsy, pneumothorax was detected on LUS in 17 patients and on CXR in 11; at 24 hours, it was identified on LUS in 21 patients and on CXR in 18 (Table 3).
In all patients with confirmed pneumothorax during the examination 3 hours post-biopsy, pneumothorax was also confirmed in the subsequent examination 24 hours after the biopsy. For detecting pneumothorax, the sensitivity of LUS was 73.9% and 91.3% at the 3- and 24-hour examinations, respectively. In contrast, the sensitivity of CXR was 47.8% and 78.3%, respectively (Table 4).
In the receiver operating characteristic curve analysis, the areas under the curves (AUCs) for LUS and CXR at the 3-hour examination were 0.870±0.049 and 0.739±0.053, respectively. At the 24-hour examination, the AUC for LUS was 0.957±0.030, while that for CXR was 0.891±0.044. A statistically significant difference was observed between the AUCs of LUS and CXR at 3 hours post-biopsy (p=0.043); however, there was no statistical significance retained at the 24-hour examination (p=0.254) (Figure 3).
Discussion
In this study, handheld ultrasonography demonstrated superior diagnostic sensitivity compared to CXR for detecting pneumothorax following invasive lung procedures. Although no significant difference in diagnostic accuracy between the two methods was observed 24 hours post-biopsy, LUS exhibited significantly greater accuracy than CXR when used shortly after the procedure (at 3 hours).
Alrajhi et al. [16] conducted a meta-analysis focusing on trauma patients and reported that LUS had a sensitivity of 90.9% for diagnosing pneumothorax, which significantly surpassed the 50.2% sensitivity of CXR. Additionally, the specificity was 98.2% for LUS compared with 99.4% for CXR [16]. Alrajab et al. [9] analyzed data from patients both with and without trauma, reporting that ultrasonography had a pooled sensitivity of 78.6% and a specificity of 98.4%, while CXR exhibited a pooled sensitivity of 39.8% and a specificity of 99.3%. In the current study, the sensitivities of LUS and CXR conducted at 3 hours post-biopsy were 73.9% and 47.8%, respectively. These results align with previous studies indicating that LUS typically demonstrates higher sensitivity compared to CXR. However, the sensitivity of LUS in our study was lower than that reported in previous research. This discrepancy may be attributed to earlier studies predominantly including trauma patients, who likely had more severe pneumothoraxes than those undergoing biopsy procedures. Moreover, while previous studies identified pneumothorax based solely on the absence of lung sliding and comet-tail signs, our study additionally required confirmation of a lung point, possibly contributing to the observed lower sensitivity.
Among the pneumothorax cases diagnosed in this study, six patients were missed by the 3-hour LUS. Of these, pneumothorax was confirmed in two patients by the 3-hour CXR. While LUS identified the absence of lung sliding in these cases, it failed to detect a lung point, a finding likely attributable to parietal emphysema. Parietal emphysema—air accumulation within or just beneath the parietal pleura could potentially interfere with accurate ultrasonographic detection [17]. This condition frequently arises from invasive procedures such as PCNB and creates a strong air-tissue interface that either reflects or absorbs ultrasound waves, thus obstructing deeper structures’ visibility, such as the lung point.
Moreover, parietal emphysema can impair normal pleural movement, resulting in the absence of lung sliding. Therefore, in cases of pneumothorax influenced by parietal emphysema, both lung sliding and the lung point may remain undetectable due to signal interference. Additionally, the increased sensitivity of CXR observed in our study compared to previous studies can be attributed to differences in patient positioning. Unlike earlier studies where CXR was typically performed while patients were in a supine position, our protocol required patients to be standing during CXR, potentially improving the detection of smaller pneumothoraxes.
Remarkably, no chest-tube insertions were necessary in this study, indicating the predominantly mild character of iatrogenic pneumothorax post-PCNB [18]. Such pneumothoraxes often resolve spontaneously without medical intervention. Furthermore, our hospital’s protocol likely contributed to this favorable outcome by promptly initiating oxygen therapy after pneumothorax detection—either on CT during the biopsy or on CXR after the procedure [19]. These preemptive measures help in preventing the progression of pneumothorax and mitigate the need for more aggressive treatments.
The significance of this research lies in its potential to provide clinicians with actionable data concerning the optimal timing for performing LUS post-invasive lung biopsy. It systematically compared the rates of pneumothorax diagnosis by LUS with those of CXR at both 3 and 24 hours following biopsy procedures. A previous study, which did not utilize a handheld ultrasonography device, found LUS to demonstrate superior sensitivity compared to CXR in diagnosing pneumothorax immediately after thoracic PCNB (80% vs. 47%) [20]. In this study, LUS showed higher sensitivity than CXR at both time intervals post-biopsy for diagnosing mild pneumothorax not requiring chest-tube placement, with a significant difference noted at the 3-hour mark. These findings indicate that performing LUS within the first 24 hours of biopsy may be more advantageous than CXR.
Furthermore, our study contributes evidence supporting the utility of LUS in promptly ruling out pneumothorax, thereby potentially enabling earlier patient discharge. Specifically, LUS at 3 hours exhibited a NPV of 91.3%, compared to 84% for CXR. When findings from both LUS and CXR were integrated, the NPV increased to 94.0%, thereby enhancing diagnostic confidence. This elevated NPV implies that LUS, particularly in conjunction with CXR, can reliably exclude pneumothorax soon after invasive procedures. Considering iatrogenic pneumothoraxes typically exhibit milder symptoms than spontaneous pneumothoraxes and are more likely to resolve spontaneously [18], patients with negative results from both imaging techniques may be safely considered for early discharge. Nonetheless, further research, including randomized controlled trials, is essential to assess the risk of progression to pneumothoraxes that necessitate intervention post-discharge.
Due to its compact design, the handheld ultrasonography device presents numerous advantages for detecting pneumothorax after invasive pulmonary examinations in clinical environments. While general ultrasonography devices are portable, their considerable weight and power requirements often hinder immediate testing. This is particularly salient given that the number of patients undergoing invasive lung biopsies greatly outnumbers those experiencing trauma-related incidents. Consequently, frequently moving heavy testing equipment or transporting patients can significantly disrupt the timely initiation of LUS after such procedures. The handheld ultrasonography device used in this study, which has an integrated battery, weighs less than 500 g and is small enough to fit into a pocket, enabling immediate clinical use. It can minimize unnecessary radiation exposure and ease the workload of medical staff. Nevertheless, the limitations associated with LUS must be recognized before its application. LUS should not be employed if lung sliding is not verifiable prior to biopsy, and lung points may remain undetected in cases of excessively large pneumothorax or parietal emphysema [17]. Furthermore, the accuracy of LUS is contingent on operator-dependency; the skill and experience of the practitioner significantly impact diagnostic reliability. A prior study investigating the learning curve of LUS among novices demonstrated that proficiency stabilized after 17 cases [21].
Our study had several limitations. First, chest CT was not used as a reference examination for pneumothorax; therefore, specificity comparisons were not feasible, as the diagnosis of pneumothorax was established by employing a composite standard of LUS and CXR during the study. The participating subjects had undergone procedures to obtain tissue samples from lung lesions. Since several participants had previously undergone multiple CT scans to confirm lung lesions, further CT scans for research purposes raised ethical concerns due to unnecessary radiation exposure. Additionally, this study primarily focused on comparing the sensitivities of LUS and CXR. Given that previous studies have reported nearly 100% specificity for each [22-24], this comparison is highly relevant. Second, PCNB was performed in a significant proportion of patients, underlining the necessity for future studies to include a broader range of procedures alongside PCNB, such as TBLB and ENB, to assess the efficacy of LUS across different biopsy techniques. Lastly, the limitation of having a single operator perform all LUS procedures could affect the generalizability of the findings to diverse clinical settings, where operator experience and variability greatly influence diagnostic accuracy. Further studies involving multiple operators and interobserver variability analysis are essential to validate these findings in various clinical contexts.
In conclusion, handheld ultrasonography may provide a quicker diagnostic alternative to conventional CXR for identifying pneumothorax following lung biopsy. Future studies should actively pursue the incorporation of LUS in monitoring post-procedural complications, potentially leading to the development of safer and more effective diagnostic protocols for lung lesions.
PDF Links
PubReader
ePub Link
XML Download
Data Sharing Statement
Full text via DOI
Print
Download Citation
