Transesophageal endoscopic ultrasound-guided tissue acquisition of lung masses: a case series with systematic review and meta-analysis

Introduction

Fine-needle aspiration (FNA) or fine-needle biopsy (FNB) guided by endoscopic ultrasound (EUS) or endobronchial ultrasound (EBUS) is an accurate, safe and minimally invasive procedure for the evaluation of mediastinal lesions in the subcarinal, paraesophageal area, aortopulmonary window, and para-aortic area. Advances in EUS and EBUS have reduced the dependence on more invasive procedures, such as video-assisted thoracoscopic surgery and mediastinoscopy, for tissue sampling in the mediastinum [1]. Worldwide, lung cancer is one of the leading causes of mortality and morbidity [2]. Tissue acquisition (TA) of lung tumors using EUS or EBUS is an emerging procedure. It can be done by EBUS-guided transbronchial needle aspiration (TBNA) via the trachea, or by EUS-FNA. The utilization of combined endobronchial and esophageal endosonography is recommended by European guidelines for diagnosing and staging lung cancer [3]. For the lesions immediately adjacent to the esophagus, TA can be achieved using FNA guided by EBUS with the scope in the esophagus (EUS-B-FNA), or directly by EUS FNA or FNB [4].

Data on the use of transesophageal EUS-guided TA from lung lesions is limited. Moreover, large comparative studies analyzing the difference in outcome using EUS-FNA and EUS-B-FNA are lacking. In this study, we report our experience of the diagnostic utility of EUS-guided sampling for lung masses, along with a systematic review and meta-analysis to obtain a summary estimate of the diagnostic yield and accuracy of EUS-guided TA from lung lesions.

Patients and methods

Results

Present series

The analysis included 14 patients (9 male; median age 60.5, range 31-76 years). All patients had prior bronchoscopic or EBUS-guided biopsy attempts that had failed because the lesion was in a difficult location. Additional samples were taken from involved mediastinal nodes in 7 (50%) patients. Table 1 shows the patients’ details along with the lesion locations and procedural details. Fig. 1 shows the details of one of the cases included in this study.

Table 1 Details of the patient and lung lesions included in the study

Figure 1 (A) High-resolution computed tomography showing a mass in right lower lobe; (B, C) 18F-FDG PET/CT showing FDG-avid right lower lobe mass without any metastasis; (D) Hypoechoic lung mass on endoscopic ultrasound with 22-G FNA needle in-situ; (E) Core biopsy specimen; (F) Microscopic findings suggestive of high-grade neuroendocrine carcinoma

18F-FDG PET/CT, fluorine-18 fluorodeoxyglucose positron emission tomography/computed tomography; FNA, fine-needle aspiration

The median size of the tumor in the longest axis was 49 (range 31-71) mm. The median number of passes was 3 (range 2-3). The EUS-FNA from lung mass was diagnostic in 13 (92.8%) patients. In one patient, the FNA sample from a lung lesion was suspicious for malignancy, and the associated subcarinal node FNA showed squamous cell carcinoma, confirming the diagnosis without requiring additional procedures. There were no reported early or late AEs following the EUS-FNA. Adenocarcinoma was the most common diagnosis (5/14), followed by squamous cell carcinoma (4/14) and small-cell carcinoma (2/14).

Systematic review and meta-analysis

Literature search and study characteristics

The search criteria yielded 2166 studies, of which 19 [10-28] were included in the meta-analysis. Fig. 2 shows the PRISMA flow diagram for study selection and inclusion. Table 2 shows the characteristics of the studies included in the meta-analysis. Among the included studies, 17 were full-text articles [10-17,20-28], while 2 were conference abstracts [18,19]. Based on the study design, 6 studies were prospective [11,13,18,24,26,27], and the remaining 13 were retrospective [10,12,14-17,19-23,25,28]. The diagnostic test was EUS-FNA/FNB in 15 studies [10-21,24,25,28] and EUS-B-FNA [22,23,26,27] in 4 studies. Ten studies used only a 22-G needle [10-12,14,16,21,22,24-26], 7 studies used a variety of needle sizes ranging from 19-25-G [15,17,18,20,23,27,28], and 2 studies did not report the type of needle used [13,19]. Supplementary Table 2 shows the details of the study quality analysis. Study quality assessment showed that 2 were of high quality [11,27], 10 were of medium quality [10,13,15,16,20-24,28], and 7 were of low quality [12,14,17-19,25,26].

Figure 2 PRISMA flowchart for study identification and selection process

TA, tissue acquisition

Table 2 Characteristics of studies included in the meta-analysis

Sample adequacy

The adequacy of the sample was reported by all 19 studies [10-28]. The pooled rate of sample adequacy was 95.4% (95% confidence interval [CI] 93.1-97.8; I²=43.0%), with low heterogeneity among the studies (Supplementary Fig. 1). On subgroup analysis, based on the type of sampling procedure, both EUS-FNA (95.4%, 95%CI 92.5-98.4; I²=42.3%) and EUS-B-FNA were found to have comparable diagnostic adequacy (95.2%, 95%CI 90.8-99.6; I²=58.9%; P=0.925).

Diagnostic accuracy

All 19 studies [10-28] reported diagnostic accuracy. The pooled diagnostic accuracy rate was 93.4% (95%CI 90.7-96.1; I²=46.0%), with low heterogeneity among the studies (Fig. 3). On subgroup analysis based on the type of sampling procedure, both EUS-FNA (93.6%, 95%CI 90.1-97.1; I²=50.3%) and EUS-B-FNA were found to have comparable diagnostic accuracy (92.6%, 95%CI 88.5-96.7; I²=30.5%; P=0.716).

Figure 3 Forest plot for pooled diagnostic accuracy with endoscopic ultrasound-guided transesophageal tissue acquisition of lung mass

CI, confidence interval

AEs

AEs directly related to the procedure were reported in 16 studies [10-17,20-25,27,28]. The reported AEs included self-limited severe chest pain, hemoptysis, para-aortic hematoma, and pneumothorax. The pooled rate of AEs with transesophageal EUS-guided TA from lung masses was 0.7% (95%CI 0.0-1.6%; I²=0.0%), without any heterogeneity among the studies (Supplementary Fig. 2).

Publication bias, sensitivity analysis, and meta-regression

Visual inspection of the funnel plot showed the presence of publication bias for diagnostic accuracy, but not diagnostic adequacy or AEs (Supplementary Fig. 3). Table 3 shows the sensitivity analysis based on the study design, type of echoendoscope used, needle size and study quality. On “leave-one-out” meta-analysis, there was no difference in the effect size of pooled diagnostic yield or pooled diagnostic accuracy. Meta-regression was not conducted as there was no significant heterogeneity regarding various outcomes.

Table 3 Summary table with sensitivity analysis

Discussion

Tissue diagnosis is essential for managing lung mass detected on cross-sectional imaging. EUS is conventionally used in staging lung cancer, but there are no recommended guidelines on EUS-guided TA for diagnosing lung masses. In this study, we reviewed our own experience of TA from lung masses and also performed a systematic review in order to carry out a critical appraisal of the available evidence on this topic. In the present analysis, the pooled sample adequacy and diagnostic accuracy were 95.4% (95%CI 93.1-97.8) and 93.4% (95%CI 90.7-96.1), respectively. Further, both EUS-FNA and EUS-B-FNA were found to have comparable adequacy and diagnostic accuracy. The present case series shows a diagnostic accuracy of 92.6%, comparable to the pooled diagnostic accuracy rate of 93.4% obtained from the meta-analysis. There were no AEs associated with EUS-FNA in the present case series, a finding consistent with the low pooled AE rate of 0.7% (95%CI 0.0-1.6%) in the meta-analysis.

For lung masses that are in peripheral locations, computed tomography (CT)-guided transthoracic needle biopsy has been found to have a higher diagnostic yield (odds ratio [OR] 0.23, 95%CI 0.13-0.42; P<0.001) and accuracy (OR 0.43, 95%CI 0.25-0.74; P=0.002), at the cost of a higher risk of complications (OR 7.27, 95%CI 5.61-9.43; P<0.001) than EBUS-TBNA [29]. For centrally located tumor masses, CT-guided biopsy has a high false-negative rate [30], apart from a greater incidence of complications such as pneumothorax needing chest tube placement [31]. EUS-guided transesophageal TA is particularly useful for centrally placed pulmonary masses, especially when the tumor invades or is adjacent to the mediastinal compartment [32]. In addition to obtaining biopsies from pulmonary mass, EUS further helps acquire additional biopsies from the mediastinum and upper abdomen simultaneously for staging purposes.

EUS-B-FNA has advantages compared to EBUS-TBNA in that it minimizes patient discomfort as the airway is not compromised [33]. However, there are other drawbacks associated with the use of the EBUS scope. The EBUS scope is shorter than the EUS scope, leading to suboptimal visualization of the left adrenal gland. The lack of a channel for air insufflation poses problems. Further, image quality is suboptimal in the EBUS scope, and the sonographic angle is narrow. Because of its narrow caliber, EBUS has lesser stability in the esophagus, and the lack of an elevator mechanism means that the needle angle cannot be altered. So, theoretically, the EUS scope fares better than EBUS in the complete staging of pulmonary masses and tissue sampling. However, on subgroup analysis based on the type of sampling procedure, our meta-analyses showed that EUS-FNA (93.6%, 95%CI 90.1-97.1) and EUS-B-FNA (92.6%, 95%CI 88.5-96.7) have comparable diagnostic accuracy (P=0.716). Combined EUS and EBUS are complementary methods for staging lung cancer [4].

The mean size of the lesion in the included studies varied from 27-68 mm, and the lesion was not a determinant of heterogeneity for either sample adequacy or diagnostic accuracy. In the study by Mangiovillano et al [28], nodule size at the cutoff of 15 mm was reported as a significant predictor of higher diagnostic accuracy. Peripherally located lesions may represent an impediment to the sample using EUS-FNA/B, which may have to be sampled using the traditional EBUS or CT-guided approach. However, EUS-FNA/B represents a potentially efficacious and safe alternative for centrally located tumors, even small-sized lesions. In a previous study of patients with centrally located lung tumors and a non-diagnostic bronchoscopy, EUS-FNA diagnosed lung cancer in 31 of 32 patients (97%) without complications [11].

The rate of AEs in the present series and meta-analysis was very low. In a systematic review by Von Bartheld et al, the complication rate for TA through endosonography for mediastinal lesions was low, at 0.14% [34]. Most of these were infectious complications in the form of mediastinitis and mediastinal abscess formation, which can be life-threatening. Necrosis in the mediastinal lymph nodes was a risk factor for infectious complications. However, during TA of lung masses, such complications are rarely encountered, as was observed in a meta-analysis of esophageal endosonography for the diagnosis of intrapulmonary tumors. Esophageal perforation, although theoretically reported, is very rare [35].

For TA, FNA needles are commonly used. In an algorithm proposed by Bang et al for needle selection, 19 G, 22 G, and 25 G needles are used for transesophageal FNA [36]. In a randomized controlled crossover trial, EUS-FNB had a superior diagnostic yield for non-pancreatic masses compared to EUS-FNA (88.2% vs. 54.5%, P=0.006), with EUS-FNB being cost-effective compared to EUS-FNA [37]. However, in that study EUS-FNB was performed using 2 passes without on-site cytopathology evaluation. In EUS-FNA, the number of passes was dictated by on-site cytopathology evaluation. On-site cytopathology evaluation is not routinely practiced in Asian countries, apart from US-based centers [38]. However, newer-generation FNB needles have superseded the utility of FNA needles, which are rarely used. A recent network meta-analysis by Gkolfakis et al showed that end-cutting FNB needles had the best performance while sampling solid pancreatic masses compared to side-cutting and FNA needles [39]. No previous studies have assessed different needles in solid lung masses, and EUS-FNB for lung mass biopsy has only been reported in one study. The most recent multicentric, retrospective study from 8 Italian centers evaluated the feasibility, accuracy, and safety of transesophageal EUS-FNA/FNB for sampling lung nodules [28]. The reported overall diagnostic accuracy was 88.9% (76.3-96.2). EUS-FNB was associated with a higher sensitivity (100% vs. 78.73%, P=0.05), diagnostic accuracy (100% vs. 78.57%, P=0.05) and sample adequacy (100% vs. 78.5%, P=0.05), compared to EUS-FNA. On multivariate analysis, nodule size >15 mm (OR 2.29, 95%CI 1.04-5.5; P=0.05) and the use of an FNB needle (OR 4.33, 95%CI 1.05-6.31; P=0.05) were significant predictors of higher diagnostic accuracy. Thus, FNB may also be the procedure of choice over FNA for solid lung lesions.

Contrast-enhanced harmonic EUS (CH-EUS) can help identify the target for EUS-FNA, with easier avoidance of anechoic areas and vessels inside the tumor. Arteries supplying lung carcinoma show late onset and a variable degree of contrast enhancement (due to their bronchial origin). This allows improved targeting of enhancing tissue compared to non-enhancing necrotic zones, which should be avoided at biopsy [40]. A recent meta-analysis reported superior sample adequacy and diagnostic accuracy of CH-EUS-FNA over standard EUS-FNA for the diagnosis of pancreatic lesions [41]. However, the lack of relevant studies prevented us evaluating the role of contrast-enhanced fine-needle aspiration (CH-EUS-FNA) for TA from lung masses, and this remains an area of future research.

There are multiple limitations to the present meta-analysis. As most studies included in the meta-analyses were retrospective, it is possible that only masses comfortably accessible by the transesophageal route had been selected. Properly conducted prospective studies on lung masses are likely to eliminate this selection bias. No data were available for direct comparison of EUS and EBUS-guided TA. The ideal suction technique also remains a topic for future studies.

To conclude, EUS-FNA/B from lung masses is a safe and effective alternative to EBUS-FNA and can be considered for centrally placed masses. EUS-FNA/B is additionally useful in sampling mediastinal lymph nodes and may help in the optimal staging of lung tumors. There is a need for future prospective studies to determine whether these findings are reproducible and to refine the criteria for recommending EUS-FNA in this setting.