We confirmed three cases of air embolism, so the incidence of air embolism is three (0.21%) of the 1,400 thoracic biopsies. Clinical data are summarized in Table 1. There were 11 cases of hypotension during or after the biopsy; however, an air embolism was found in the aorta, left ventricle, pulmonary vein, or coronary artery on the CT scan (Figs. 1A, 1B, 1C, 1D, and 1E) obtained during or after biopsy in three cases. All three patients underwent biopsy with the core needle. The estimated causes of hypotension of the remaining eight cases are pneumothorax (n = 3), pain (n = 2), hemothorax (n = 1), and unknown (n = 2).

No large pulmonary vein on the needle track was shown on the CT scans obtained during the biopsy. There was normal lung tissue in the biopsy specimen in all three cases, but the pulmonary venous wall was obtained (Fig. 1C) only in case A2 (cases with air embolism are designated with an A; those with needle track insertion are designated with an N). Because the center of the tumor tends to be necrotic, we usually attempt to take the specimen from the peripheral portion of the lesion. Thus, the specimen might include normal lung tissue even in a case in which the lesion was larger than the size of the notch.

The length between the pleura and the lesion was 0–2 cm (mean, 1.2 cm). Hypotension occurred during the biopsy in two patients (cases A2 and A3) who immediately needed resuscitation. Although an air embolism was retrospectively confirmed in the left ventricle on the CT scan obtained during the biopsy, the hypotension was revealed not during but after the biopsy in case A1. Although there were no fatal cases, left hemiplegia occurred after resuscitation in case A3. Brain CT and MRI were obtained after the event, but no apparent air embolism or infarction was noted in this particular case. Although hyperbaric oxygen therapy was not performed, the symptoms gradually disappeared 1 month later.

The malignancy was pathologically confirmed in 891 of the 1,400 thoracic cases, and we confirmed four cases of needle track implantation among them. The number of patients and the time interval between CT-guided biopsy and the last follow-up chest CT were as follows: 178 patients, 0–2 months; 25 patients, 3–5 months; 66 patients, 6–11 months; 117 patients, 12–23 months; and 505 patients, 2 years or more). We excluded the 178 patients with 0–2 months from the total number of patients with reliable follow-up. Thus, the incidence of needle track implantation was four (0.56%) of 713 cases. Clinical data are summarized in Table 2.

The size of the implantation on the CT scans ranged from 2.5 to 5.6 cm (mean, 3.5 cm) when found on follow-up CT scans. The interval between the biopsy and detection of the needle track implantation ranged from 4 to 7 months (mean, 5.6 months). All of these four intrathoracic lesions were surgically resected (thoracotomy, cases N1, N2, and N4 and video-assisted thoracic surgery, case N3) after CT-guided biopsy. There was no macroscopic pleural seeding, and intraoperative pleural lavage cytology was negative at the time of surgery.

The stage of primary lung cancer of case N1 was T1N0M0 at surgery, and needle track implantation had occurred without other recurrence at the time of follow-up CT. Needle track implantation with local pleural invasion (Figs. 2A and 2B), contralateral pleural dissemination, and hilar node metastasis had developed at the time of follow-up CT in case N2. Not only local pleural invasion of needle track implantation but also massive ipsilateral pleural dissemination had occurred at the time of follow-up CT in case N3. The stage of the primary lung cancer of case N4 was T2N1M0 at surgery, and needle track implantation with local pleural invasion, ipsilateral pleural dissemination, and adrenal and abdominal lymph node metastasis had occurred at the time of follow-up CT. No patient received chemotherapy after the surgery except the patient in case N3 who had renal cell carcinoma and received interferon-γ. Case N1 has survived without other metastasis. The patients in cases N2, N3, and N4 were dead 2, 9, and 5 months, respectively, after the detection of needle track implantation.

There were multiple metastases in three of the four cases in which needle track implantation was found. Needle track implantation was removed in case N1 without any further metastasis. The location of the tip of the 18-gauge coaxial needle was the same location as the implantation (Figs. 2A and 2B) in all of four patients.

The reported incidence of air embolism in surveys of large numbers of cases was two (0.07%) of 2,726 cases in 1976 by Sinner [3], one (0.018%) of 5,444 cases in 2002 by Richardson et al. [1], six (0.06%) of 9,783 cases in 2006 by Tomiyama et al. [4], and four (0.4%) of 1,010 cases in 2007 by Hiraki et al. [2]. However, the report by Sinner was published more than 30 years ago using aspiration needles. According to the multiinstitutional questionnaire studies of Richardson et al. and Tomiyama et al., the mean number of biopsies performed per center was 30 [1] and 21 [4] per year, which was fewer than our cases. In addition, the biopsy method, including needle type, size, and so on, was not unified across the different hospitals in these studies. The study by Hiraki et al. was the only recent report from a single institution using the core biopsy needle. Although only one of the four patients with an air embolism was symptomatic in the study by Hiraki et al., the incidence of air embolism in our institution is close to their result. They recommended that the entire thorax be scanned for a survey of possible air embolisms after the biopsy. The entire thorax scan after each needle pass seems to be better in terms of the early detection of an air embolism; however, it would be complicated in practicality.

To clarify the mechanism of the air embolism, we pathologically reviewed biopsy specimens obtained from patients with a confirmed air embolism. Normal lung tissue was confirmed in all specimens, and a relatively remarkable pulmonary venous wall was noted in one case (case A2). Therefore, there was a possibility that the air entered the pulmonary vein through the injured vein in this particular case. However, the large pulmonary vein was not seen on the needle track on the CT scan in three patients including this particular patient. We could not find the mechanism causing the air embolism in the other two cases using the biopsy specimen. We always plan the needle path for the biopsy not to puncture the large vessels, especially the pulmonary vein; however, it is practically impossible to avoid cutting normal lung tissue, including some part of the small venous wall. In addition, when we reviewed the previously reported cases with an air embolism (Table 3), not only the core needle but also other types of needles (mostly aspiration needles) caused air embolism.

The longer the distance between the pleura and the lesion, the more the needle damages the lung parenchyma, including the pulmonary vein and the bronchiole, and it may form a transient communication between them. The distance between the lesion and the pleura in our patients with air embolism was 0–2.0 cm (mean, 1.2 cm), so it was not longer than the distance (0.3–4.0 cm; mean, 2.1 cm) in the reported thoracic biopsy cases [4]. Therefore, we think the length between the pleura and the lesion does not play an important role in causing air embolism.

We summarize both the previously reported 16 symptomatic cases with detailed clinical information of air embolism [2, 7–21] and our three cases in Table 3. Hemiplegia was noted on the right (n = 2) or left (n = 8) side in the 10 cases with neurologic symptoms. Brain CT was performed in 13 cases in which an air embolism in the cerebral artery was confirmed in nine patients; right (n = 7), left (n = 1), and multiple (n = 1). The positions of the 11 patients with left hemiparesis or air embolism in the right cerebral artery were supine (n = 5), left decubitus (n = 4), or prone (n = 2). The positions of the two patients with right hemiplegia or air embolism in the left cerebral artery were prone (n = 2). Because the right brachial artery branches first from the aortic arch, we assumed that the air tended to enter the right carotid artery, irrespective of the patient's position, but it might enter the left carotid artery only when the patient is in the prone position.

Air embolism was confirmed in either the right (n = 4) or left (n = 3) coronary artery in seven cases. The patient position on the table at the time of the biopsy was supine (n = 3) or left decubitus (n = 1) in the four cases with air embolism in the right coronary artery. In contrast, the position was right decubitus (n = 1) or prone (n = 1) in the two cases with an air embolism in the left coronary artery. The right coronary artery is located ventrally when the patient is in the supine position, and the left coronary artery is located ventrally when the patient is in the right decubitus position. Therefore, we assumed that the air tended to enter the coronary artery located ventrally at the time of biopsy because air is lighter than blood.

An air embolism was also confirmed in the ascending artery (n = 8), left pulmonary vein (n = 3), left ventricle (n = 6), and left atrium (n = 3) on CT scans.

According to previous reports, five patients died, with hyperbaric oxygen therapy performed on one patient (Table 3). In contrast, hyperbaric oxygen therapy was performed in four of the 14 cases in which the patient survived. Although one patient immediately died on the CT table [11], it is recommended that the patient with an air embolism be sent to the hyperbaric oxygen unit if possible. However, hyperbaric oxygen therapy is not always available in all institutions, and there were survivors who did not undergo this therapy. Cardiopulmonary resuscitation, endotracheal intubation, and oxygen are necessary for immediate cardiac life support. According to Muth and Shank [22], it is recommended that patients with air embolism should be placed in the flat supine position, and heparin is beneficial but corticosteroids remain controversial for the treatment of air embolism.

When we began to perform CT-guided biopsy 15 years ago, we placed the cuff around the patient's arm to measure blood pressure intermittently during the procedure. Now, we always use a pulse oximeter for continuous monitoring, in addition to the sphygmomanometer in all patients.

Ayar et al. [5] estimated the incidence of needle track implantation as 0.012% on the basis of their multiinstitutional survey study. According to their study, the five institutions reported one (0.071%) of 1,400 cases to one (0.2%) of 500 cases as the incidence of needle track implantation. Kim et al. [6] reported eight (0.2%) of 4,365 patients developed needle track implantation related to percutaneous fine-needle aspiration thoracic biopsy. Matsuguma et al. [23] reported that pleural recurrence (n = 5, 7.5%) or needle track implantation (n = 1, 1.5%) was observed in 66 patients who underwent percutaneous needle biopsy preoperatively. This is a higher incidence rate of needle track implantation than in our study; however, they used an 18-gauge biopsy needle, whereas we used a 20- or 21-gauge biopsy needle. Therefore, the incidence of needle track implantation may depend on the size of the biopsy needle.

The total number of the biopsies used to calculate the incidence of needle track implantation in the previous studies was not the total “malignancy proven” cases but the total number of the biopsy cases including both benign and malignant lesions. Thus, we thought that the reported incidence was less than the actual incidence. In addition, it is necessary to follow up cases after biopsy to observe whether needle track implantation has occurred. Therefore, we think the incidence of needle track implantation with follow-up in a single institution is more accurate.

The incidence of needle track implantation after the biopsy in cases of liver lesions was 2.7%, according to a meta analysis [24]. Percutaneous ethanol injection therapy [25] and radiofrequency ablation for hepatocellular carcinoma (HCC) [26] also cause needle track implantation. The incidence of needle track implantation of thoracic malignancies was smaller than that of HCC, which may be due to the unknown difference between lung tumor and HCC in terms of biologic behavior of malignancy.

Although one lesion was an “undifferentiated” carcinoma (adenosquamous carcinoma), the other lesion was a “common type” (papillary adenocarcinoma) in our particular cases with primary lung cancer. The other two lesions were lung metastasis and were already advanced cancer cases. Previous reports also suggested that all types of pulmonary malignancies could result in needle track implantation [5, 6].

According to Ayar et al. [5], the previously reported size of needle track implantation was 8 mm (biopsy; implantation interval, 2 weeks) to 7 cm (biopsy; implantation interval, 5 months). Therefore, the size of the tumors reported is not unusual. The previously reported interval of several months between the time of the biopsy and the development of needle track implantation was not long. The mean was 2.6 months (range, 2–12 months) in one study [5], and the mean was 7 months (range, 2–16 months) [6] in another. In other words, tumor growth was very fast in cases with needle track implantation. Therefore, we should make close observations of patients who have undergone biopsy for at least 1 year, even if the lung lesion was surgically removed. When we review follow-up CT in patients who underwent CT-guided percutaneous biopsy, we examine not only the intrathoracic lesions but also the thoracic wall to find any needle track implantation as early as possible.

Because the location of needle track implantation was almost the same as the tip of the coaxial needle, we assumed that the tumor cells attaching to the biopsy needle were shed at the tip of the coaxial needle during withdrawal. If the coaxial needle is placed adjacent to the surface of the lesion in the lung, the semiautomatic biopsy needle is further advanced into the lesion and the biopsy is performed. The incidence of needle track implantation in the chest wall would be diminished. However, the placement of a needle may cause an air embolism, so we did not place a coaxial needle in the lung.

We have not found the proper techniques for complete prevention of needle track implantation. Thus, when the benefit of biopsy is superior to the risk of implantation in terms of diagnosis, a biopsy should be recommended.

There were patients who became hypotensive during or after the biopsy; however, air embolism was not always detected on the CT scan obtained at the biopsy. When the lung lesion was located at the apex, CT was not usually performed at the level of the ascending aorta or left ventricle during the biopsy. Thus, an air embolism could be missed in those patients.

We did not examine the resected lung specimen with respect to the incidence of needle track implantation. Therefore, a tiny implantation in the lung tissue would be missed. The real incidence of needle track implantation including lung parenchyma may be more than that we have reported here.

In conclusion, the incidence of air embolism with hypotension and needle track implantation complicating percutaneous CT-guided thoracic biopsy was thought to be rare; however, the incidence was 0.21% and 0.56%, respectively, in our retrospective study—more frequent than the previously reported rates. Because we cannot prevent air embolism and needle track implantation completely, it is necessary to prepare for immediate resuscitation during biopsy and make close observation of patients for at least 1 year after the biopsy, respectively.