Evaluation of mediastinoscopy in mediastinal lymph node staging for non-small-cell lung cancer

Abstract

OBJECTIVES

The purpose of this study was to assess the quality of video-assisted cervical mediastinoscopy (VACM) in the staging of non-small-cell lung cancer (NSCLC) at the Antwerp University Hospital with a focus on test effectiveness indicators, morbidity and unforeseen pN2 results.

METHODS

All consecutive VACM workups of cases of NSCLC performed between January 2010 and December 2015 were included to assess overall test quality and effectiveness. Quality assurance was performed in accordance with the recommendations of the European Society of Gastrointestinal Endoscopy and European Society of Thoracic Surgeons (ESTS) where appropriate.

RESULTS

A total of 168 video-assisted cervical mediastinoscopies were included. A total of 91.7% of the procedures were performed in accordance with the ESTS guideline. An unforeseen pN2 staging was identified in 10 anatomical lung resections (8.6%). Statistical analysis showed no significant association between VACM performed in accordance with the ESTS guideline and the presence of pN2 positive lymph nodes [χ² (1) = 0.61; P = 0.57] and no association between VACM performed in accordance with the ESTS guideline and overall futile thoracotomy [χ² (1) = 0.76; P = 0.50]. Calculations revealed a sensitivity of 81.8 [95% confidence interval (CI) 69.1–90.9], specificity of 100%, positive predictive value of 100%, negative predictive value of 91.9% (95% CI 86.6–95.2) and diagnostic accuracy of 94.1% (95% CI 89.33–97.11).

CONCLUSIONS

Overall, 91.7% of the VACM were performed in accordance with the ESTS guideline. This process resulted in a sensitivity of 81.8%, a negative predictive value of 91.9% and an unforeseen pN2 rate of 8.6%.

Lung cancer remains a challenging disease, with a total of 8472 new cases of non-small-cell lung cancer (NSCLC) observed in Belgium in 2017 [1].

INTRODUCTION

Lung cancer remains a challenging disease, with a total of 8472 new cases of non-small-cell lung cancer (NSCLC) observed in Belgium in 2017 [1]. In the workup of a suspected case of NSCLC, computed tomography (CT) and 18F-labelled fluoro-2-deoxyglucose positron emission tomography (PET) are often combined to assess tumour location, size and metastases. However, radiological imaging does not provide histological information and lacks sufficient diagnostic accuracy. Combined with the fact that lymph node (N)-staging plays a pivotal role in planning the therapeutic approach, invasive staging is often necessary [2]. Cervical mediastinoscopy, with lymph node identification as described by Naruke, remains the gold standard for N-staging [3–5]. The cervical mediastinoscopy technique was first described by Carlens [6] in 1959. Due to the invasive nature of cervical mediastinoscopy and the advancement of endosonographic techniques, the revised European Society of Thoracic Surgeons (ESTS) guideline of 2014 advises endobronchial ultrasound (EBUS), if possible, combined with endoscopic transoesophageal ultrasound (EUS) as the initial approach to N-staging. If endosonographic staging is negative for N2 or N3 lymph nodes, the guideline states that this result needs to be confirmed with a surgical biopsy via video-assisted cervical mediastinoscopy (VACM) [7]. However, VACM is associated with possible major complications including trachea rupture, oesophagus rupture, severe haemorrhage or paresis of the laryngeal recurrent nerves.

The purpose of this study was to assess the quality of VACM in the staging of NSCLC in the Antwerp University Hospital, with a focus on test effectiveness indicators, morbidity and unforeseen pN2 results.

MATERIALS AND METHODS

The present study was approved by the ethics review board of the Antwerp University Hospital (Doc nr: 20/18/243; dated 11 May 2020). Because this was an observational retrospective study, the need for informed consent was waived.

Data were collected retrospectively. All consecutive VACM images for the workup of NSCLC performed between January 2010 and December 2015 were included. This study period was selected to ensure overall staging according to the seventh edition of the tumour, node, metastasis (TNM) classification [8]. The decision to perform a VACM was made during a multidisciplinary team meeting and was based on one of the following indications: cN1–3 observed on PET-CT scans, central location of a primary tumour, peripheral location of a primary tumour >3 cm, negative findings of a primary tumour on 18F-labeled fluoro-2-deoxyglucose-PET scan, diagnostic tissue sampling and elective restaging after neoadjuvant therapy.

All mediastinal lymph nodes with a short axis greater than 1 cm on CT scans were categorized as positive. The primary tumour was labelled as a central tumour when located in the inner third of the hemithorax or when in contact with structures of the hilum. If the multidisciplinary team had advised an EBUS or EUS, data relating to these procedures were also included. The quality of the endosonographic staging was compared to the quality standards of the European Society of Gastrointestinal Endoscopy (ESGE) guideline, created in cooperation with the European Respiratory Society and the ESTS, published in 2015. For an EBUS to have been performed in accordance with the ESGE guideline, stations 7, 4L and 4R should be biopsied at least 3 times. In the case of EUS, biopsies should also be taken from station 8.

The quality of the VACM staging was compared to the quality standards of the revised ESTS guidelines [4]. According to the ESTS guideline, VACM must include biopsies of at least 7, 4L, 4R, and, if present, 2L and 2R. Lymph node stations 5 and 6 were never biopsied during VACM.

The VACMs were performed by 3 surgeons (J.M.H.H., P.L., P.E.Y.V.S.). Complications were categorized according to the Clavien–Dindo classification [9]. Anatomical resection was labelled futile in the following cases: pN2, pT4 with R1 resection, pM1, small-cell lung cancer, diagnostic resection or benign tumour [4, 7]. During the anatomical resection, at least a lobe-specific systematic lymph node dissection was performed.

Statistical analyses were performed with SPSS 26.0 (IBM SPSS Statistics for Windows, Version 26.0. IBM Corp., Armonk, NY, USA). Continuous variables are expressed as median with range. Categorical variables are expressed in percentages. Data were compared with χ² test, Fisher’s exact test and the Kruskal–Wallis test where applicable. P-values smaller than 0.05 were considered statistically significant. The sensitivity, specificity, positive and negative predictive values and diagnostic accuracy were calculated according to formulas by Galen and Gambino [10].

RESULTS

A total of 168 VACM were included. The median age was 65 years (range 40–83 years) at the time of VACM and 69.0% of the procedures were done on male patients. Other demographic data are displayed in Table 1. Median preoperative forced expiratory volume in 1 s was 88.5% (range 35–149%), and median corrected diffusion capacity was 79.0 mmol/min.kPa.L (range 26–120 mmol/min.kPa.L).

Table 1:

Demographics

Variables	Results (n = 168), n (%)
Active smoker	93 (55.4)
Smoking cessation <15 years	14 (8.3)
Non-smoker	61 (36.3)
Chronic obstructive pulmonary disease	45 (26.8)
Arterial hypertension	46 (27.4)
Diabetes mellitus	14 (8.3)
Cardiovascular history	39 (23.3)
Renal insufficiency	6 (3.6)
Malignancy history	28 (16.7)

Endosonographic mediastinal staging took place using EBUS in 59 out of 168 patients (35.1%); in 2 of the 59 cases (3%), EBUS was combined with EUS. Of the 59 endosonographic procedures, 32% (n = 19) were performed according to the ESGE guideline. The ESGE guideline was not adhered to (n = 40) as follows: not all lymph node stations were biopsied systematically in 29 procedures (73%); insufficient tissue was procured in 6 procedures (15%); and the PET-positive lymph node was not biopsied in 5 procedures (13%). Of the 59 cases, 9 (15%) had a positive N2–3 staging and 2 (3%) were suspected to be positive N2–3, yet not enough material was obtained to provide a definite result. Hence, these 2 also had a VACM. Of these 2, an N3 stage was confirmed in 1 case. The other case, however, had negative results for both VACM and anatomical resection. A total of 48 out of 59 endosonographic procedures (81%) had an N0–1 staging; these patients then had VACM. Nonetheless, VACM revealed an N2–3 stage in 15 out of 48 (31%). In 6 of these 15 procedures (40%), VACM identified a positive lymph node that had initially been marked negative on endosonographic staging. Of the remaining 9 out of 15 (60%) procedures, in 1 case the lymph node was biopsied, but not enough material was obtained for diagnosis, whereas in the other 8 cases (53%), the positive lymph node had not been biopsied on EBUS. No endosonographic-related complications or deaths were documented. Since the revised ESTS guidelines were published in 2014, a total of 55 VACM were performed, of which 33% (n = 18) were preceded by endosonographic staging. In 18 endosonographic procedures, 11 (61%) of the EBUS/EUS procedures were performed in accordance with the ESGE guideline.

Indications for the 168 VACM were as follows: cN1–3 on PET-CT scans in 53.6% (n = 90), peripheral tumour >3 cm in 13.7% (n = 23), centrally localized tumour in 11.9% (n = 20), diagnostic tissue sampling in 7.1% (n = 12), no primary tumour found on 18F-labeled fluoro-2-deoxyglucose-PET scans in 1.2% (n = 2) and elective restaging after neoadjuvant therapy in 12.5% (n = 21). Neoadjuvant therapy preceding the VACM comprised chemotherapy alone in 19.6% (n = 33), sequential chemoradiotherapy in 2.4% (n = 4) and concomitant chemoradiotherapy in 1.8% (n = 3). Additional relevant preoperative variables are listed in Table 2. A median of 4 lymph node stations (range 1–7 stations) was biopsied. Complete distribution of lymph node biopsies is displayed in Table 3.

Table 2:

Preoperative data

Variables	Results (n = 168), n (%)
Tumour localization
Right upper lobe	48 (28.6)
Right middle lobe	14 (8.3)
Right lower lobe	28 (16.7)
Left upper lobe	57 (33.9)
Left lower lobe	21 (12.5)
Premediastinoscopy histological diagnosis
Adenocarcinoma	28 (22.0)
Squamous cell carcinoma	37 (27.0)
Neuroendocrine tumour	2 (1.2)
Non-small-cell lung cancer of unknown type	3 (1.8)
Unknown	98 (58.3)
cTNM
T1a	24 (14.3)
T1b	29 (17.3)
T2a	43 (25.6)
T2b	24 (14.3)
T3	35 (20.8)
T4	13 (7.7)
N0	61 (36.3)
N1	12 (7.1)
N2	76 (45.2)
N3	19 (11.3)
M0	160 (95.2)
M1a	3 (1.8)
M1b	5 (3.0)

cTNM: clinical tumour, node and metastasis.

Table 3:

Number of biopsied lymph node stations

Variables	Results (n = 168), n (%)
1 station	3 (1.8)
2 stations	2 (1.2)
3 stations	15 (8.9)
4 stations	65 (38.7)
5 stations	55 (32.7)
6 stations	27 (16.1)
7 stations	1 (0.7)

VACM (n = 168) was performed in accordance with the ESTS guidelines in 91.7% of procedures (n = 154). Reasons for not complying with the ESTS guidelines were: not all required lymph node stations were biopsied in 7 procedures (4.2%); the risk of a biopsy was deemed too high due to extensive fibrosis in 4 procedures (2.4%); and 1 patient had unsafe anatomy (0.6%). The VACM consisted of a redo VACM in 9 procedures (5.4%).

A VACM-related major complication (Clavien–Dindo 3) occurred in 1 out of 168 procedures (0.6%); it was a deep surgical site infection requiring surgical drainage. Minor complications (Clavien–Dindo 1–2) occurred in 9 procedures (5.4%). Of the minor complications, 4 were recurrent laryngeal nerve lesions (2.4%). The median hospital stay was 2 days (range 1–9 days). There were no VACM-related deaths.

A total of 45 of the 168 patients who had VACM had a positive N2–N3 staging. Table 4 displays their histological diagnosis and distribution. Of these 45 cases, 15 (33%) had negative results on earlier endosonographic staging procedures. In the case of N0 staging, anatomical lung resection was performed (n = 116). An unforeseen pN2 staging was identified in 10 out of 116 resections (8.6%); of these positive lymph nodes, 8 (80%) had been biopsied according to the ESVS guideline during VACM and were found to be negative. The remaining 2 nodes were not located in the superior mediastinum (lymph node stations 6 and 9) and were not biopsied during VACM. A futile resection occurred in 12 cases (10.3%) because of positive pN2 in 10 procedures (83%), pT4 with an R1 resection in 1 patient (8%) and an pM1a staging in 1 patient (8%). Pathological TNM staging is listed in Table 5.

Table 4:

N2–3 positive lymph nodes on mediastinoscopy

Variables	Results
N-staging	(n = 45), n (%)
N2	35 (78)
N3	10 (22)
Number of positive stations	(n = 45), n (%)
1	24 (53)
2	12 (27)
3	4 (9)
4	3 (7)
5	2 (4)
Total number of positive stations	82
Location of positive stations	(n = 82), n (%)
2R	16 (20)
2L	7 (9)
4R	17 (21)
4L	15 (18)
7	27 (33)
Histologically positive stations	(n = 45), n (%)
Adenocarcinoma	24 (53)
Squamous cell carcinoma	7 (16)
Neuroendocrine tumour	2 (4)
Non-small-cell lung cancer of unknown type	12 (27)

Table 5:

pTNM staging following anatomical resection

pTNM	Results (n = 116), n (%)
T0	7 (6.0)
T1a	19 (16.4)
T1b	15 (12.9)
T2a	30 (25.9)
T2b	16 (13.8)
T3	24 (20.7)
T4	5 (4.3)
N0	82 (70.7)
N1	24 (20.7)
N2	10 (8.6)
M0	115 (99.1)
M1a	1 (0.9)

pTNM: pathological tumour, node and metastasis.

There was no statistically significant association between the presence of unforeseen pN2 positive lymph nodes and the number of lymph nodes biopsied during VACM [χ² (6) = 6.10; P = 0.42] nor when analysing the association between performing VACM (according to the ESTS guidelines) and the presence of pN2 positive lymph nodes [χ² (1) = 0.61; P = 0.57]. Neither was there a significant association between performing VACM (according to the ESTS guidelines) and overall futile thoracotomy [χ² (1) = 0.76; P = 0.50].

Applying the formulas by Galen, the VACM in the study population had an overall sensitivity, specificity and diagnostic accuracy of 81.8% [95% confidence interval (CI) 69.1–90.9], 100% and 94.1% (95% CI 89.3–97.1), respectively. In addition, calculations showed a negative predictive value of 91.9% (95% CI 86.6–95.2) and a positive predictive value of 100% (Table 6). Subgroup analysis of the population before 2014 (n = 113) showed a sensitivity, specificity and diagnostic accuracy of 85.4% (95% CI 70.8–94.4), 100% and 94.7% (95% CI 88.8–98.0), respectively. A negative predictive value of 92.3% (95% CI 85.1–96.2) was observed along with a positive predictive value of 100%. In contrast, subgroup analysis of the population from 2014 onwards (n = 55) had a sensitivity, specificity and diagnostic accuracy of 71.4% (95% CI 41.9–91.6), 100% and 93% (95% CI 82.4–98.0), respectively. A negative predictive value of 91.1% (95% CI 81.7–95.9) was observed along with a positive predictive value of 100%.

Table 6:

Diagnostic test evaluation values

	Total video-assisted mediastinoscopies (n = 168)
	Positive	Negative
Actual outcome
Positive	True positive: 45	False negative: 10
Negative	False positive: 0	True negative: 113

DISCUSSION

Our department follows the ESTS guidelines closely, with VACM performed according to the ESTS guidelines in 91.7% of the procedures. This result reflects a sensitivity of 81.8%, negative predictive value of 91.9% and diagnostic accuracy of 94.1%. Inherent in the status of gold standard, the VACM has a specificity and positive predictive value of 100%. These results fall within the sensitivity range of 70–95% reported in the literature [11–13].

We used the seventh edition of the TNM classification for lung cancer from 2010 to 2015. Consequently, the 2014 revised ESTS guidelines for implementation of endosonographic staging preceding VACM only cover a small portion of the study population [4]. Of the subpopulation from 2014 onwards, 61% of the endosonographic procedures were performed according to the ESGE guidelines. This situation contrasts with the 33% correctly performed procedures in the population through the end of 2013. From 2014 onwards, all PET-positive lymph nodes were biopsied using endosonographic staging; 13% were missed in the period prior to 2014. Therefore, the assumption that the procedures performed before the ESTS update would focus more on PET-positive lymph nodes rather than undertake a standard approach seems to be erroneous. The fact that adherence to the guideline improved is a positive trend; however, even further adherence to the guideline should be the goal. If primary mediastinal staging is indicated, our centre favours an initial correct endosonographic staging in accordance with the ESGE guideline for quality assurance [7]. In the case of an N0–1 stage, we adhere to the ESTS guidelines and believe that VACM should be performed to confirm operability [4]. As in our study, out of 48 negative results on EBUS/EUS scans, VACM revealed a N2–3 stage in 15 cases (31%). Of these 15 cases, 6 positive lymph nodes (40%) were found to be negative on earlier endosonographic staging. This result underscores the fact that, in our experience, VACM confirmation is still required. The position of VACM is also currently being investigated in the MEDIAST trial, a multicentre parallel randomized non-inferiority trial comparing endosonographic staging of NSCLC with and without VACM [14].

Minor complications (Clavien–Dindo 1–2) occurred in 9 out of 168 VACM procedures (5.4%), with a specific complication, lesions to the recurrent laryngeal nerve, in 4 cases (2.4%). A major complication (Clavien–Dindo 3) occurred in only 1 patient (0.6%), i.e. a deep surgical site wound infection requiring surgical drainage. The literature reports a VACM-related major morbidity of 0.6–3.7% and a recurrent laryngeal nerve lesion in up to 6% of cases [15, 16]. A videolaryngoscopic examination to assess vocal cord paralysis and diagnose recurrent laryngeal nerve palsy was performed only in symptomatic patients. Our series found no complications related to endosonographic staging. The literature does confirm lower morbidity for endosonographic staging compared to VACM of up to 1.44% [17]. As shown in the recent systematic review and meta-analysis by Sehgal et al. [18], there is a significantly higher occurrence of major complications with mediastinoscopy in comparison to endosonographic staging with an absolute risk difference complication rate of 6.8% (95% CI 4.3–9.8). Although the risks of serious complications are significantly higher, high-level evidence is growing that the sensitivities of both VACM and EBUS/EUS are similar when viewed separately; however, the combined sensitivity remains superior with a sensitivity of nearly 94% [19–21].

Unforeseen N2 staging was found in 10 resection specimens (8.6%), 8 (80%) of which were biopsied via VACM before the anatomical resection. The positive lymph nodes contained micrometastases and were probably missed due to a sampling error. The 2 positive lymph nodes that were not initially biopsied were stations 6 and 9. Retrospectively, both lymph nodes were PET-positive. A correct staging including EUS could have identified station 9 as positive and could thus have prevented a futile anatomical resection.

Zhang et al. [22] confirmed a cumulative sensitivity of 86% when combining EBUS and EUS. In our study, however, only 2 cases (3.4%) of the 59 endosonographic procedures had a combined EBUS and EUS examination. The meta-analysis by Bousema et al. [23] demonstrated that even in the case of a full negative diagnostic work-up including EBUS, EUS and mediastinoscopy, unforeseen pN2 occurred in 7–9.3% of cases. However, the presence of pN2 might not lead to decreased overall survival or progression-free survival. The European Organization for Research and Treatment of Cancer trial compared patients with proven N2 (stage IIIa-N2) treated with induction chemotherapy who received either radiotherapy or surgical resection; results showed similar overall survival and increased progression-free survival rates for the surgical arm [24]. The patients receiving concurrent chemoradiotherapy in the Essen-Paris-Tübingen (ESPATUE) trial were randomized to a definitive concurrent boost of chemoradiotherapy or surgical resection. In this study, overall survival and progression-free survival were similar, supporting the hypothesis that surgical resection is a suitable strategy in this population [25].

Overall, N2-disease is considered a contraindication for primary surgery; therefore, multimodal therapy should be recommended if N2 disease is present. However, the role of surgery within this multimodal therapy is still up for debate [4]. Due to the heterogeneity of the involved lymph nodes, a new N-staging is addressed in the eighth edition of the TNM classification with a further division of N2 lymph nodes in N2a1, N2a2 and N2b [26, 27]. The essence would be to further divide the N2 metastases to identify atypical lymphogenic metastases. N2a1 refers to a single N2 positive lymph node without N1 disease, formally referred to as a ‘skip N2’ disease. In N2a2, again, a single N2 positive lymph node is present; however, N1 disease is also present. In the case of N2b, multiple positive lymph node stations would be involved. As shown by Cerfolio and Bryant [28], N2a2 (single level N2, with N1 involvement) disease showed a significantly better 5-year survival compared to N2b (multilevel) (40% vs 25%, respectively; P = 0.028). Similar survival rates for unforeseen N2 disease were reported by Obiols et al. (40%) and Lee et al. (33.5%) [29, 30]. Hence, the questions arise if surgical biopsies via VACM are truly required to exclude N2 disease or if endosonographic staging with a possibly higher unforeseen N2 rate is acceptable.

Limitations

A limitation of this study is its retrospective nature. However, observational data were only rarely missing; hence we do believe that the study accurately reflects our current practice. Even so, under-reporting of minor complications cannot be excluded. A second limitation of the study may be its sample size. The study period was selected to guarantee staging according to the seventh edition of the TNM classification [8]. As the study demonstrates, there was close adherence to the ESTS guidelines during this period in our centre, and we did not identify any significant outcome associations [4]. Inclusion of centres with a lower adherence rate would increase the overall sample size and might reveal significant differences in outcome when comparing different, less ESGE and ESTS adherent, lymph node staging strategies. A final limitation is that our centre is a tertiary academic hospital; hence, procedures are occasionally partially performed by surgical trainees. The exact involvement of the surgical trainee during the procedure is, however, only rarely mentioned in the operative reports, and all procedures are performed under the direct supervision of a staff member.

CONCLUSIONS

With a high ESTS guideline adherence of 91.7%, we report the following VACM test values: sensitivity (81.8), specificity (100%), positive predictive value (100%), negative predictive value (91.9%) and diagnostic accuracy (94.1%). Our study confirms the need for VACM confirmation after negative endosonographic staging. However, with the advancements in standardized endosonographic staging in the workup of NSCLC, the position of VACM needs to be re-evaluated with high-quality trials.

Abbreviations

CI

Confidence interval

CT

Computed tomography

EBUS

Endobronchial ultrasound

ESGE

European Society of Gastrointestinal Endoscopy

ESTS

European Society of Thoracic Surgeons

EUS

Endoscopic (transoesophageal) ultrasound

NSCLC

Non-small-cell lung cancer

PET

Positron emission tomography

TNM

Tumour, node, metastasis

VACM

Video-assisted cervical mediastinoscopy

 

Conflict of interest: none declared.

Author contributions

Ian Diebels: Data curation; Formal analysis; Investigation; Methodology; Software; Writing—original draft. Jeroen M.H. Hendriks: Supervision; Writing—review & editing. Jan P. Van Meerbeeck: Supervision; Writing—review & editing. Patrick Lauwers: Writing—review & editing. Annelies Janssens: Writing—review & editing. Suresh K. Yogeswaran: Visualization; Writing—review & editing. Paul E.Y. Van Schil: Conceptualization; Methodology; Supervision; Writing—review & editing.

Reviewer information

Interactive CardioVascular and Thoracic Surgery thanks Giuseppe Cardillo, Marcin Zielinski and the other, anonymous reviewer(s) for their contribution to the peer review process of this article.