The lung is usually considered poorly accessible by sonography. However, transthoracic sonography has been proven to be a useful method for the evaluation of pleural effusions, allowing detection of small volumes of pleural fluid [1, 2]. Transthoracic sonography has emerged as a diagnostic tool in various diffuse and focal pulmonary diseases [2-5] and in pneumothorax and hydropneumothorax [6-8]. Sonographic images of the lung are composed exclusively of artifacts because air prevents transmission of the ultrasound beam, but the artifacts in air-containing lungs differ from those induced by air in the pleural space [6]. In particular, in healthy lungs, respiration-dependent movement of the visceral pleura and lung surface with respect to the parietal pleura and chest wall can be easily visualized with real-time transthoracic sonography. This characteristic is known as lung sliding or the gliding sign [8-10]. At the boundary between the pleura and the ventilated lung, intensive bandlike reverberation echoes (comet-tail artifacts) are evoked during breathing movements [6, 9]. Comet-tail artifacts are generally sporadic in healthy lung and become more numerous in diffuse parenchymal diseases [2, 9]. The presence of air in the pleural space prevents sonographic visualization of visceral pleural movements, and the gliding sign and comet-tail artifacts disappear [2, 9, 10]. In addition, pleural air generates reverberation artifacts that form parallel horizontal echoic lines characterized by artifactual immobility during breathing movements. These artifacts (frozen echoes) are well documented with M-mode imaging [6, 9, 10]. With these main findings, bedside transthoracic sonography is widely used to detect pneumothorax in critically ill patients [11-14]. Conversely, the usefulness of transthoracic sonography in excluding and diagnosing postinterventional pneumothorax has scarcely been investigated. The few reports published are case reports [15] or trials in which a small number of patients were enrolled and conflicting results were obtained [10, 16]. This prospective study was planned to evaluate the accuracy of transthoracic sonography in the detection of pneumothorax after transthoracic sonographically guided biopsy of peripheral pulmonary lesions.

Postbiopsy pneumothorax occurred in eight (2.8%) of 285 patients. In seven cases, pneumothorax was asymptomatic and resolved spontaneously in 1 or 2 days. In one case, pneumothorax necessitated chest tube insertion and drainage for 3 days. In five cases, pneumothorax occurred immediately after biopsy; in three cases, it was documented 3 hours later. Transthoracic sonography correctly depicted all cases of pneumothorax and correctly excluded the presence of pneumothorax in 277 patients. Chest radiographs depicted seven cases of pneumothorax. In one case, transthoracic sonography showed disappearance of the lung lesion immediately after biopsy and depicted the lung point approximately 3 cm from the lesion, but chest radiographs did not depict pneumothorax immediately or 3 hours after biopsy. In this patient, thoracic CT findings confirmed the presence of a small pneumothorax limited to the surroundings of the biopsy site.

In all patients with pneumothorax, transthoracic sonography showed absence of lung sliding and comet-tail artifacts and presence of horizontal reverberation artifacts. The lung point was identified in seven patients, whereas it was not identified in the patient who needed chest tube insertion.

The sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy of transthoracic sonography were 100%. Chest radiography had a sensitivity of 87.5%, specificity of 100%, positive predictive value of 100%, negative predictive value of 99.6%, and overall accuracy of 99.6%. The 95% CIs for sensitivity and specificity of transthoracic sonography were 63-100% and 89.5-100%, respectively; those of chest radiography were 47.3-99.6% and 89.5-100%. The 95% CIs of the differences in sensitivity, negative predictive value, and overall accuracy between transthoracic sonography and chest radiography were -10 to 35%, -0.1 to 0.9%, and -0.1 to 0.9%.

Transthoracic sonographic monitoring of postbiopsy pneumothorax correctly showed resolution of pneumothorax in all cases, depicting reappearance of lung sliding and comet-tail artifacts and disappearance of horizontal reverberation artifacts and frozen echoes. There was 100% concordance with chest radiographic findings

Transthoracic needle biopsy is a well-established technique for diagnosing pulmonary lesions, and transthoracic sonographic guidance has been found as effective as CT guidance for lesions in contact with pleura [17-19]. Pneumothorax is the most common complication of transthoracic sonographically guided pulmonary biopsy, the frequency ranging from 1.6% to 5% [18, 20-22]. Chest radiography is routinely performed 3 hours after the procedure to assess development of pneumothorax, but small quantities of intrapleural air can escape detection [10]. CT is the most sensitive method for diagnosing pneumothorax, but it cannot be considered the method of choice in clinical practice because of the high exposure to ionizing radiation, considerable cost, and difficulties arising in transporting patients [10, 14]. Therefore there is increasing interest in alternative techniques in routine postinterventional exclusion of pneumothorax [10, 16] and detection of pneumothorax in trauma and critical illness [11-14].

Sonograms of the lung are exclusively composed of artifacts, but the artifacts generated by healthy lung can differ from those of several pulmonary pathologic conditions [2-5, 9]. Results of some studies [6, 8] have suggested that transthoracic sonography may be useful in the detection of pneumothorax. In particular, disappearance of lung sliding and comet-tail artifacts and detection of horizontal reverberation artifacts have been shown to be reliable signs of pneumothorax. More recent prospective studies [10, 14] confirmed these findings, and we have documented them in 100% of cases of pneumothorax in our study group.

Transthoracic sonography has been reported to have a high negative predictive value in detection of pneumothorax. The positive predictive value may be slightly lower, because a number of coexisting situations can hamper sonographic examination and cause false-positive findings [14]. In particular, bullous emphysema, pachypleuritis with pleural adhesions, and phrenic paralysis may impair lung excursion and prevent visualization of lung sliding, simulating the presence of pneumothorax [10, 14, 23, 24]. In our study, patients were enrolled after prebiopsy sonographic assessment of lung sliding and comettail artifacts, and no false-positive findings were observed after transthoracic sonographically guided pulmonary biopsy. Consequently, to exclude the presence of coexisting conditions that could hamper the diagnostic accuracy of transthoracic sonography, we recommend that preliminary assessment of these signs be performed immediately before patients undergo transthoracic sonographically guided interventional procedures. In this limited number of cases (in our experience, 1%), chest radiography should be preferred to transthoracic sonography.

Transthoracic sonography is considered a promising technique for detection and exclusion of pneumothorax in critically ill and trauma patients [11-14, 25]. Conversely, there are only limited and conflicting data available on the use of transthoracic sonography in the diagnosis of postinterventional pneumothorax. In an early case study, transthoracic sonography depicted two cases of pneumothorax after transthoracic sonographically guided biopsy of peripheral lung lesions [15]. In a subsequent study, however, transthoracic sonography did not depict five of 13 cases of postinterventional pneumothorax [16]. In that study, however, sonographic examination was restricted to the area around the needle entry site, and that methodologic limitation is likely to have hampered the diagnostic accuracy of transthoracic sonography. When transthoracic sonographic examination was extended to the entire anterior chest wall, the sensitivity and specificity of transthoracic sonography were 80% and 94% in an evaluation of pneumothorax after fluoroscopically guided lung biopsy [23]. Reissig and Kroegel [10] studied the accuracy of transthoracic sonography in detection of pneumothorax in 53 patients undergoing transbronchial biopsy or chest tube placement. Conducting systematic sonographic examination of the anterior and posterior intercostal spaces on the side of biopsy or tube thoracostomy, those authors found 100% sensitivity and 100% specificity of transthoracic sonography in the diagnosis or exclusion of postinterventional pneumothorax. Our study design was similar to that of Reissig and Kroegel, and our findings confirm their results but with a substantially larger number of patients. Moreover, our results suggest that transthoracic sonography can be helpful in documenting resolution of pneumothorax, replacing chest radiographic monitoring and avoiding patient exposure to ionizing radiation.

Despite its sensitivity in detection of pneumothorax, transthoracic sonography is not considered a reliable tool for estimating the volume of pneumothorax [14, 26]. Although the size of pneumothorax can be roughly approximated by assessment of the lung point, the depth of pulmonary collapse cannot be evaluated. Chest radiography is necessary to determine the extent of pneumothorax when the lung point is identified far from the biopsy site [10]. For this reason, we did not attempt sonographic estimation of the size of postbiopsy pneumothorax. An interesting finding, however, was that we identified the lung point in all cases of pneumothorax except the one necessitating chest tube drainage. Although anecdotal, this observation also has been reported by other authors [23] and suggests that unsuccessful detection of the lung point may be a sign of deep pulmonary collapse and be an indication for mandatory prompt performance of chest radiography. Further confirmation with a larger number of patients is needed to substantiate such a hypothesis.

Our study had two limitations. First, CT was performed only when discrepancies were found between transthoracic sonographic and chest radiographic findings. Consequently, it is possible, although unlikely, that small, asymptomatic cases of pneumothorax were not detected with both transthoracic sonography and chest radiography and that the diagnostic accuracy of transthoracic sonography was overestimated but without consequence for the patients' clinical conditions. The second limitation was caused by the low prevalence of pneumothorax after transthoracic sonographically guided pulmonary biopsy [18, 20-22]. In our series the prevalence was 2.8%. It follows that despite the large number of patients enrolled, the low number of cases of pneumothorax detected may make the estimate of sensitivity questionable and decrease the ability to generalize the data. This limitation makes evaluation of sensitivity in larger series of cases of pneumothorax mandatory before any definitive conclusion can be drawn.

Despite the limitations, according to several reports [10-14, 23, 25], our results suggest that transthoracic sonography, performed by trained physicians, is an accurate tool for exclusion of the presence of pneumothorax. It is a safe and noninvasive bedside method, is at least as accurate as chest radiography, and does not expose patients to ionizing radiation. Moreover, the technique has been reported to have good interobserver reproducibility when the operators are skilled [23].

If further prospective studies confirm our findings, transthoracic sonography should be considered the method of choice in clinical practice for excluding, diagnosing, and monitoring pneumothorax after transthoracic sonographically guided interventional procedures. Chest radiography may be needed only for assessment of the extent of pulmonary collapse after pneumothorax has been diagnosed with transthoracic sonography and the lung point has been identified far from the biopsy site or in the presence of a discrepancy between transthoracic sonographic findings and the patient's clinical condition.

Address correspondence to S. Sartori ([email protected]).