Abstract
Objectives: The aim of the study was to assess the diagnostic yield of the endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) in restaging of the non-small-cell lung cancer (NSCLC) patients after neo-adjuvant therapy. Methods: In a consecutive group of NSCLC patients with pathologically confirmed N2 disease, who underwent neo-adjuvant chemotherapy, EBUS-TBNA was performed. All patients with negative EBUS-TBNA underwent subsequently the transcervical extended bilateral mediastinal lymphadenectomy (TEMLA) as a confirmatory test. Results: A total of 61 patients underwent restaging EBUS-TBNA between 1 June 2007 and 31 December 2008. There were 85 mediastinal lymph nodes biopsied (stations: 2R – 2, 4R – 24, 2L – 1, 4L – 18 and 7 – 40). EBUS-TBNA revealed metastatic lymph node involvement in 18 of 61 patients (30%) and in 22 of 85 biopsies (26%). In 43 patients with negative or uncertain EBUS-TBNA, who underwent subsequent TEMLA, metastatic nodes were diagnosed in nine patients (15%) – in seven (12%) in stations accessible for EBUS-TBNA (stations: 2R – 1, 4R – 5, 7 – 4) and in two (3%) in station not accessible for EBUS-TBNA (station: 5 – 2). The false-negative results of biopsies were found only in small nodes (5.8 ± 2.8 mm × 7.5 ± 2 mm). Moreover, all positive N2 nodes diagnosed by TEMLA contained only small metastatic deposits. There were three of 61 (5%) patients with false-positive results of biopsies in stations: 4R – 1, 4L – 1, and 7 – 2. A diagnostic sensitivity, specificity, accuracy, positive predictive value (PPV) and negative predictive value (NPV) of the restaging EBUS-TBNA was 67% (95% confidence interval (CI) – 65–90), 86% (95% CI – 82–95), 80%, 91% (95% CI – 80–100) and 78% (95% CI – 73–93), respectively. No complications of EBUS-TBNA were observed. Conclusions: EBUS-TBNA is an effective and safe technique for mediastinal restaging in NSCLC patients, and after the data presented in our study, in patients with negative results of EBUS-TBNA, a surgical restaging of the mediastinum might not be mandatory.
1 Introduction
Real-time endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is a recently introduced method of biopsy of the mediastinal, hilar and interlobar lymph nodes [1–5]. Although widely accepted in non-small-cell lung cancer (NSCLC) staging, little is known about the role of EBUS in lung cancer restaging after neo-adjuvant therapy. The actual data show high specificity and accuracy of this method, but low negative predictive value (NPV) and sensitivity [6,7]. It has been suggested that negative results of EBUS-TBNA in restaging should be confirmed by surgical staging before thoracotomy. There are currently no accepted standards regarding mediastinal restaging, and many strategies, based on radiological, minimally invasive and surgical techniques are advocated [8].
According to the recent European Society of Thoracic Surgeons’ and American College of Chest Physicians’ guidelines, minimally invasive procedures, such as EBUS-TBNA and ultrasound-guided endoscopic needle aspiration (EUS-NA) may be alternatively used for mediastinal staging, but mediastinoscopy or remediastinoscopy should be preferably reserved for restaging [8–10]. A low positive predictive value (PPV) of positron emission tomography-computed tomography (PET-CT) and a low NPV of biopsy techniques and mediastinoscopy does not give the answer as to which technique is appropriate for restaging. In any of the published studies assessing the value of the EBUS-TBNA in NSCLC restaging, its sensitivity and NPV were confirmed by the mean of the bilateral mediastinal lymphadenectomy – generally, mediastinoscopy or thoracotomy were used. Mediastinoscopy enables only access to 5 out of 13 mediastinal nodal stations and the extent of mediastinal exploration at remediastinoscopy is even less; the same is true for exploration at thoracotomy. So, some percentage of the negative results of the EBUS-TBNA must have been missed, causing a bias in the reported results (Fig. 1 ).
Restaging EBUS-TBNA in station 7.
Restaging EBUS-TBNA in station 7.
2 Material and methods
2.1 Clinical question
What is the real sensitivity, specificity, accuracy, PPV and NPV of EBUS-TBNA in the NSCLC restaging after neo-adjuvant therapy, assessed using the bilateral mediastinal lymphadenectomy as the confirmatory test?
2.2 Design
The study design was a prospective cohort diagnostic study.
2.3 Location
The study was carried out at the Department of Thoracic Surgery, Pulmonary Hospital, Zakopane, Poland.
2.4 Patients
Inclusion and exclusion criteria for the patients are listed below:
Inclusion criteria: A group of consecutive NSCLC patients: (1) Clinical stage IIIA–IIIB confirmed only by EBUS-TBNA, (2) after three to five series of neo-adjuvant chemotherapy (cisplatin + vinorelbine), (3) with enlarged or normal mediastinal lymph nodes on CT scans – defined by radiologists as having stable disease or partial response after induction chemotherapy and (4) general condition enabling appropriate pulmonary resection.
Exclusion criteria: (1) no response after induction chemotherapy – a progression of radiological or endobronchial lesions and (2) lack of patient’s consent.
2.5 Intervention
Prior to the procedure the CT scans were carefully analysed. EBUS-TBNA was performed under local anaesthesia and intravenous sedation (fentanyl 0.05–0.1 mg, midazolam 1–5 mg), using the BF-UC160F-OL8 videobronchoscope (Olympus Medical Systems Corporation, Tokyo, Japan). The videobronchoscope is 6.9 mm wide, has a 2-mm working channel, the 35° optical system and the EU-C60 7.5 MHz ultrasound processor, enabling 20–50 mm depth tissue imaging. As the ultrasound bronchoscope is not designed for detailed assessment of the bronchial tree, the examination was preceded by the standard videobronchoscopy. For the biopsy, a cytological 22G needle with guide wire and marking facilitating its visualisation on the ultrasound image was used (NA-201SX-4022, Olympus Medical Systems Corporation, Tokyo, Japan).
EBUS-TBNA of detectable lymph nodes ≥5 mm on the short axis were performed (criterion of feasibility of lymph node biopsy according to Herth et al. [4]); first of all, enlarged nodes or those considered suspect based on the shape or echogenicity were biopsied. If there were no bigger nodes visible, even those >3 mm were punctured. All the biopsies were performed through the macroscopically normal bronchial wall. The number of biopsied nodal stations in one patient was 1–4.
The cytological smear was performed and fixed using 96% ethanol. Standard haematoxilin–eosin staining was used and the specimens were sent to the cytological laboratory.
In patients with negative results of EBUS-TBNA, transcervical extended mediastinal lymphadenectomy (TEMLA) was performed. TEMLA includes bilateral dissection of all the mediastinal lymph nodes, except for station 9. The use of a special retractor, elevating the sternum, enables access to the mediastinal structures and safe dissection of lymph nodes. The technique of TEMLA is described in detail elsewhere [11,12]. The video presenting this technique is available at: www.mp.pl/download/wmv/temla.wmv.
In patients with negative results of TEMLA, an appropriate pulmonary resection with dissection of the mediastinum was performed. The extent of the mediastinal dissection corresponded to the systematic lymph node dissection. However, due to the completeness of lymphadenectomy with the TEMLA technique, generally no nodes were found at thoracotomy.
The Mountain–Dresler lymph node classification was used [13].
2.6 Statistical analysis
Statistical calculations were carried out using Statistica™ software (Statsoft Inc., Tulsa, OK, USA).
Summary statistics were expressed as mean (M) and standard deviation (SD). The sensitivity, specificity, accuracy, PPV and NPV were calculated using the standard definitions. The McNemar test was used, where appropriate, for paired comparisons. For patient-based analyses, confidence intervals were calculated to 95% using standard formulae. For biopsy-based analyses, confidence intervals were based on asymptotic normality of maximum-likelihood estimators and delta method for logit function. The type I error was set at 0.05 for all analyses.
3 Results
There were 49 men and 12 women in the mean age 59.3 ± 7.5 years (range: 44–77 years). In these 61 patients, 85 EBUS-TBNA were performed, and the diagnosis was established in 18 (30%) of them.
No complications of EBUS-TBNA were observed. A small bleeding from the place of biopsy was not considered as a complication. Among these 85 biopsies, there were 40 subcarinal node (station 7) biopsies, 24 right lower paratracheal node (station 4R) biopsies, 18 left lower paratracheal node (station 4L) biopsies, two right upper paratracheal node (station 2R) biopsies and one left upper paratracheal node (2L) biopsy. The mean diameter of the biopsied nodes was 13.3 ± 8.1 mm in the long axis and 9.5 ± 6.4 mm in the short axis. In 24 patients (39%) biopsy of more than one node was performed.
EBUS-TBNA was technically successful in 75 cases (88%), and in 22 out of 85 biopsies (26%) the persistent nodal metastases were detected.
In 18 true positive patients (30%), metastatic involvement of the lymph node was confirmed in 22 nodal stations (in three patients more than one node was involved). The numbers of metastatic nodes in particular stations were as follows: station 7 – 11, 4R – 5, 4L – 5 and 2R – 1.
In 31 patients (51%), the result of mediastinal lymph node biopsy was true negative. The number of these true negative biopsies was 49: 23 in station 7, 13 in station 4R, 12 in station 4L and one in station 2L. In this group, the cytological diagnosis of benign, reactive lymph node enlargement was subsequently confirmed by the histological examination of the TEMLA operative specimen. In addition, in 25 patients (41%) with negative result of the TEMLA, mediastinal lymph node dissection during thoracotomy was performed and no metastatic nodes were found in them. The next six patients (10%) after the TEMLA did not undergo lung resection: five out of them had a significant impairment of pulmonary function tests that made lung resection impossible and one out of them refused the second surgery. In the majority of patients, after the TEMLA, asymptomatic widening of the mediastinum on chest X-ray was seen and in one patient transitory left recurrent nerve palsy was observed.
In nine patients (15%) the result of EBUS-TBNA was false negative. In seven patients (12%), TEMLA revealed metastases in stations accessible for EBUS-TBNA (stations: 7 – 4, 4R – 5, 2R – 1) and in the next two patients (3%) in station 5, not accessible for EBUS-TBNA. In all nodal stations with false-negative results of the EBUS-TBNA, the extent of metastatic involvement was low (mean 48 ± 24% of the nodes, Table 1 ). All the false-negative results of biopsies were obtained in small nodes 5.8 ± 2.8 mm × 7.5 ± 2 mm. There were three patients (5%) with false-positive results of biopsies in stations: 4R – 1, 4L – 1 and 7 – 2.
Patients with false-negative results of restaging EBUS-TBNA verified by the TEMLA.
Patients with false-negative results of restaging EBUS-TBNA verified by the TEMLA.
The overall sensitivity of EBUS-TBNA calculated per patient basis was 67% (95% confidence interval (CI) – 65–90), specificity – 86% (95% CI – 82–95), accuracy – 80%, PPV – 91% (95% CI – 80–100) and NPV – 78% (95% CI – 73–93). However, if calculated for the nodal stations accessible for EBUS-TBNA, these figures were: sensitivity – 72% (95% CI – 67–90), specificity – 86% (95% CI – 82–95), accuracy – 84%, PPV – 92% (95% CI – 80–100) and NPV – 83% (95% CI – 76–97).
The diagnostic yield of the EBUS-TBNA calculated per station is presented in Table 2 .
Results of EBUS-TBNA in NSCLC restaging in different stations of mediastinal lymph nodes calculated by numbers of biopsies.
Results of EBUS-TBNA in NSCLC restaging in different stations of mediastinal lymph nodes calculated by numbers of biopsies.
The prevalence of mediastinal lymph node metastases in the present study was 44%.
EBUS-TBNA accurately changed the clinical stage in 41 patients (67%): down-staged from N2 to N0 disease in 39 patients and from N2 to N1 disease in two patients (Table 3 ).
The radiological (R), EBUS-TBNA-based (EBUS) and TEMLA-based (TEMLA) clinical TNM staging in 61 patients.
The radiological (R), EBUS-TBNA-based (EBUS) and TEMLA-based (TEMLA) clinical TNM staging in 61 patients.
The mean time of EBUS-TBNA was 15.4 ± 6.5 min (range 10–23) and the mean time of the restaging TEMLA was 102 ± 18.3 min (range 78–125).
4 Discussion
In fact, despite technical difficulties due to mediastinal fibrosis, remediastinoscopy is considered a valuable tool in the restaging of NSCLC after neo-adjuvant therapy [8,14]. The reported sensitivity of remediastinoscopy is 29–76%, specificity and PPV – 100% and accuracy – 60–88% [14–17]. Invasive techniques are indicated despite the encouraging results supported with the use of PET-CT imaging, mainly because of its low PPV [8,16,18]. In some articles, the sensitivity and accuracy of PET-CT may be even higher than for remediastinoscopy: 77% and 83%, respectively, most probably because remediastinoscopy performed after thorough primary mediastinoscopy and neo-adjuvant therapy is technically difficult and less extensive [16].
If the restaging biopsies (EBUS-TBNA, EUS-NA or mediastinoscopy) confirm persistent N2 disease, non-surgical treatment is indicated in most patients [8,15]. The use of the EBUS-TBNA in NSCLC restaging is gaining increasing acceptance, due to its efficiency and low invasiveness, particularly as the initial procedure. Sensitivity of EBUS-TBNA was reported to be 76%, specificity and PPV – 100% and accuracy – 77% [7]. Many authors believe that, in the near future, this technique may become the gold standard not only in staging, but may also replace, in many cases, mediastinoscopy and other invasive techniques in restaging [1,2,7,9,19]. As the risk of complications related to the EBUS-TBNA and EUS-NA is very low, these procedures may be performed in the outpatient settings [3,20]. Endobronchial ultrasound enables very accurate localisation of the extrabronchial structures, including vessels (using the power Doppler imaging) and lymph nodes. An interpretation of visualised mediastinal or hilar structures after induction chemo- or radiotherapy is much more difficult than in standard staging images, and requires much more experience from the endoscopist. Gaining skill and experience in the endobronchial ultrasound imaging and performing the biopsy in restaging is more time-consuming and operator-dependent than in standard staging performed by EBUS. It is hardly possible to distinguish between suspected region for metastases and post-inflammatory adhesions, degenerative changes and fibrosis after oncological treatment. This may also be the reason of the technical failure in 12% of the biopsies. But even when not characteristic enough and in a relatively remote position from the bronchial wall, the suspected region can be biopsied by use of 10- to 40-mm long needles, mainly because of real-time imaging.
Whereas the sensitivity of the EBUS-TBNA in staging amounts to 88–95%, the reported range of its sensitivity in restaging is narrower and amounts to 75–78% [2,4,7,19,21]. Our results confirm this data: the overall sensitivity of the EBUS-TBNA in restaging was 67%, but if calculated for the nodal stations accessible for EBUS-TBNA it was 72% and not significantly different (p = 0.42).
EBUS-TBNA enables the biopsy of stations 2R, 2L, 3P, 4R, 4L and 7 (also N1 stations 10R, 10L, 11R and 11L that are inaccessible for other invasive techniques, which is a unique advantage of EBUS-TBNA). Imaging and biopsy of paratracheal nodes (particularly on the left side) is technically more difficult than the subcarinal nodes and in our study, the diagnostic yield of the EBUS-TBNA of subcarinal station was also higher than in paratracheal stations, although the difference did not reach the level of significance (p = 0.08) (Table 2). In our experience, a posterior and inferior part of the station 4L and 7 is more difficult to visualise and to biopsy. This may be the main reason for some of the false-negative results. In such circumstances, EUS-NA may be a valuable supplement, as it is very useful in assessment of the stations 4L, 7, 8 and 9. Using both EBUS-TBNA with EUS-NA or EUS-NA with mediastinoscopy enables more accurate assessment of the mediastinum and increases the diagnostic yield in staging, but there is no data for such combination in restaging [20,22].
The rate of false-negative biopsies was only 12%, and only in small nodes 7.5 ± 2 mm × 5.8 ± 2.8 mm. Moreover, in all nodal stations with false-negative results of EBUS-TBNA, the extent of metastatic involvement was low (mean 48 ± 24% of the nodes, Table 1). In most of these cases the extent of metastatic nodal involvement was limited, which was confirmed by the pathological examination of the TEMLA specimen (Table 1). According to the largest series published to date, the NPV of the EBUS-TBNA in staging is 85–96% [8,9,19] while between the two articles on TBNA and EBUS-TBNA in restaging, only one reported the NPV which amounted to only 20% [6,7]. In our series the overall NPV of the EBUS-TBNA in restaging was – 78% (95% CI – 73–93) but if calculated for the nodal stations accessible for EBUS-TBNA, it was as high as 83% (95% CI – 76–97) (difference not significant; p = 0.54), despite use of a very accurate confirmatory test – the TEMLA [23]. In our study, the majority (87%) of these false-negative results happened to be nodal sampling error rather than detection error. This also may be caused by the fact that primary staging was performed only by EBUS-TBNA and no massive adhesions or fibrosis were observed in the mediastinum after induction chemotherapy.
We considered as false positive the cases which were assessed ‘uncertain’ by the pathologist, and in whom the histological examination of the TEMLA specimen did not confirm metastatic deposits. Therefore, it should be probably considered the sequel to neo-adjuvant therapy.
If these results in restaging are confirmed in other series, perhaps in patients with negative results of EBUS-TBNA and additionally of EUS-NA, the omitting of surgical restaging procedures may well be justified, particularly if the PET-CT is also negative [24,25].
5 Conclusions
EBUS-TBNA is an effective and safe technique for mediastinal restaging in NSCLC patients and, after the data presented in our study, in patients with negative results of EBUS-TBNA, a surgical diagnostic exploration of the mediastinum might not be mandatory.
