Extended cervical mediastinoscopy: mature results of a clinical protocol for staging bronchogenic carcinoma of the left lung^†

Abstract

OBJECTIVE

The objective of this study is to evaluate the accuracy of extended cervical mediastinoscopy (ECM) in the staging of bronchogenic carcinoma (BC) of the left lung based on our updated experience.

METHODS

From 1998 to 2003, 89 patients underwent routine ECM for staging of BC of the left lung. In 2004, routine positron emission tomography (PET) was included in our staging protocol and ECM was reserved for those with positive mediastinal or hilar PET images, large lymph nodes on computed tomography (CT) scan or central tumours. Following this protocol, from 2004 to 2010, we performed 132 selective ECM. ECM was considered positive when metastatic nodes or tumour involvement directly in the subaortic or para-aortic regions was confirmed pathologically. Patients with negative ECM underwent subsequent thoracotomy for tumour resection and systematic nodal dissection (SND).

RESULTS

Two hundred and twenty-one ECMs were performed from 1998 to 2010 (89 routine and 132 selective). In the routine ECM protocol, four cases were positive and thoracotomy was contraindicated. The remaining 85 patients were operated and five had nodal disease in subaortic (LN5) or para-aortic (LN6) stations. In the selective ECM protocol (n = 188), 132 patients underwent ECM and in 19 it was positive; the remaining 113 patients underwent thoracotomy and SND found involved LN5 or LN6 in six patients; the other 56 patients underwent direct thoracotomy and four had positive LN5 or LN6. Sensitivity, specificity, positive predictive value, negative predictive value and accuracy of ECM were 0.67, 1, 1, 0.94 and 0.95, respectively. The staging values of routine/selective ECM protocols were 0.44/0.65, 1/1, 1/1, 0.94/0.94 and 0.94/0.95, respectively.

CONCLUSIONS

Selective ECM protocol according to CT and PET findings has high negative predictive value and accuracy. Therefore, its selective use is recommended because it saves around 30% ECM without decreasing staging values of the current protocol.

INTRODUCTION

The importance of assessing the involvement of mediastinal lymph nodes in lung cancer is well-established, determining prognosis and therapeutic options. Bronchogenic carcinoma (BC) of the left lung can metastasize to nodes of the aorto-pulmonary window, especially if it is located in the upper lobe and hilum. In order to surgically explore the subaortic and para-aortic nodes, a left anterior mediastinotomy was traditionally needed [1]. These nodal stations can also be reached by video-assisted thoracoscopic surgery (VATS) [2–4], but there are no specific reports on that indication. Extended cervical mediastinoscopy (ECM), a technique described by Specht in 1965 [5] and popularized by Ginsberg [6] for staging BC of the left lung, allows the assessment of para-aortic and subaortic nodal stations through the same incision of the standard cervical mediastinoscopy (SCM).

This paper describes our updated experience [7] in this technique comparing two different protocols in two periods: in the first period, from 1998 to 2003, ECM was performed routinely, regardless of the results of computed tomography (CT); and in the second period, from 2004 to 2010, when positron emission tomography (PET) was routinely indicated for clinical staging, ECM was performed selectively, according to the results of CT and PET.

MATERIAL AND METHODS

From 1998 to 2010, 221 ECM were performed in patients with suspected or diagnosed BC of the left lung. We have compared two different periods with some differences in the staging protocols, one of them with routine ECM and the other with selective ECM depending on the results of CT and PET.

From 1998 to 2003, the first staging protocol included bronchoscopy, CT of the chest and upper abdomen, and bone scan; CT of the brain was reserved for symptomatic patients or in locally advanced tumours. Surgical exploration of the mediastinum was performed routinely (SCM for tumours of the right lung, and SCM with ECM for tumours of the left lung), regardless of the findings on CT scan. According to this protocol, 89 patients underwent ECM to assess the involvement of para-aortic and subaortic nodes.

Since 2004, with the introduction of PET in our staging protocol, surgical exploration of the mediastinum (SEM) was selectively performed according to the results of CT and PET. Bone scan was not performed routinely anymore. Therefore, SEM was performed in these situations: in case of enlarged lymph nodes on CT scan (over 1 cm in the short axis), in central tumours or in mediastinal/hilar abnormal uptake on PET scan. According to this protocol, 132 patients underwent ECM, and 56 patients underwent thoracotomy directly.

In both periods, operability was similarly assessed: medical history and physical examination, complete blood analysis, electrocardiogram, pulmonary function test and ventilation–perfusion lung scan in patients with FEV1 of 2000 ml or less and tumours requiring pneumonectomy, or with FEV1 of 1500 ml or less and tumours requiring lobectomy.

The surgical exploration of the mediastinum started with a mediastinoscopy. In this procedure, right and left paratracheal nodes and subcarinal nodes were biopsied or removed. In case of macroscopically suspicious nodal involvement, a sample of the node was sent for frozen section. In those cases with confirmation of malignancy, no more biopsies were carried out. However, in patients with left lung tumours, if mediastinoscopy did not show malignancy, the exploration was completed with an ECM to explore the para-aortic and subaortic nodal stations, which cannot be reached by a standard mediastinoscopy.

ECM is performed through the same incision used for mediastinoscopy. Blunt digital dissection between the innominate artery and left carotid artery is performed, moving above the aortic arch and under the left innominate vein. This manoeuvre allows the introduction of the mediastinoscope obliquely over the aortic arch to reach the subaortic nodal station. If the mediastinoscope is slightly rotated medially, the para-aortic nodes can be reached [6, 7].

ECM was considered positive when either tumour metastases in the lymph nodes or direct tumour involvement in the subaortic and para-aortic stations was pathologically confirmed. Patients with positive ECM either received definitive chemoradiotherapy or induction chemotherapy or chemoradiotherapy and further restaging if there was no evidence of disease progression. Patients with negative mediastinoscopy and ECM underwent thoracotomy for lung resection and systematic nodal dissection (SND). SND was considered the gold standard to compare the negative results of ECM. Pathological findings were reviewed and staging values (sensitivity, specificity, positive predictive value, negative predictive value and diagnostic accuracy) were calculated using the standard formulas. Chi-square test was used for group comparison when appropriate. A p-value of <0.05 was considered significant. Confidence intervals were calculated using the CIA (Confidence Interval Analysis, v. 1.0) software package (Martin J. Gardiner & British Medical Journal, 1989).

RESULTS

From 1998 to 2010, 221 ECM were performed in patients with diagnosed or suspected carcinoma of the left lung (89 ECM were performed routinely and 132 selectively). There were 206 men and 15 women with an average age of 65 years (range: 44–83 years). The clinical characteristics of patients in both periods are shown in Table 1.

In the routine ECM protocol, 89 patients underwent ECM. It was positive in four and thoracotomy was contraindicated. The remaining 85 patients with negative ECM underwent thoracotomy for lung resection and SND. In five patients, SND showed nodal involvement in subaortic or para-aortic stations. These cases were considered false-negative results of ECM.

In the group of selective ECM, 132 out of 188 patients with diagnosed or suspected left lung cancer underwent ECM according to the results on CT and PET scan, and the remaining 56 underwent direct thoracotomy. ECM was positive in 19 patients and thoracotomy was contraindicated. The remaining 113 patients with negative ECM underwent thoracotomy with SND. In six cases SND showed nodal involvement in the subaortic or para-aortic stations. Finally, in the 56 patients who underwent direct thoracotomy, SND showed N2 disease in subaortic or para-ortic stations in four. These cases were considered false-negative of the protocol.

The staging values of the technique and the protocol are shown in Tables 2 and 3, respectively.

Complication rate was 2.3% with two cases of mild mediastinitis, treated with superficial wound drainage and antibiotics; one ventricular fibrillation treated with intraoperative defibrillation; one superficial wound infection treated with wound drainage; and one haemorrhage that was solved with compression. There was no mortality associated to the procedure in this series.

DISCUSSION

Mediastinal staging of BC of the left lung includes the exploration of right and left paratracheal, subcarinal, para-aortic and subaortic lymph nodes. To assess the aortopulmonary window, traditionally an anterior mediastinotomy (Chamberlain's procedure) was performed in addition to SCM. In 1965, Specht described the ECM, in which, using the same incision of the SCM, an exploration of the aorto-pulmonary window was possible [5]. In 1987, Ginsberg popularized this technique for staging purposes [6] and we perform ECM according to his description. After digital dissection, a passage is created to advance the mediastinoscope over the aortic arch, between the innominate artery and the left carotid artery, under the left innominate vein. In our first patients, ECM was combined with parasternal mediastinotomy in order to certify the correct placement of the mediastinoscope. Other authors have described some modification to this technique, in which the dissection is made in the retrosternal space, between the anterior surface of the innominate vein and the posterior surface of the sternum. However, this variation does not seem to influence the results [8, 9]. Another surgical procedure to stage para-aortic and subaortic nodes is VATS. This technique can also assess the intrathoracic spread and evaluate the primary lung tumour, but there are few reports on the specific indication of nodal staging [2–4].

In the past few years, new non-invasive techniques have been described and developed. Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) and endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) have demonstrated high accuracy for mediastinal staging, especially if EBUS and EUS are combined and if a systematic exploration is performed [10]. The ESTS guidelines for preoperative lymph node staging recommend a non-invasive technique as first mediastinal staging test, if available. However, if this test shows a negative result, a surgical mediastinal procedure is required, owing to the low negative predictive value of these techniques [11]. Moreover, the para-aortic and subaortic stations are more difficult to reach using EUS-FNA owing to the intervening pulmonary artery and aorta and some authors have described a lower negative predictive value in cases of left lung cancer [12].

In the present series, we have updated our experience in this staging procedure [7]. The comparative results of both periods are shown in Table 4. Despite the incorporation of new staff members in the most recent period, the staging values of the technique are very similar. These results show that, after an appropriate teaching period, this technique can be reliably performed. Regarding the patient's characteristics, there are no statistical differences in both periods except on the type of lung resection performed. There was a greater percentage of pneumonectomies from 1998 to 2004 in comparison with the last period, 31.7 vs. 9.4%, respectively. This fact shows that the current tendency in most thoracic surgery services is to try to preserve lung function and to avoid the higher morbidity of pneumonectomies.

Since ECM was performed selectively, we have observed that staging values of the protocol remain similar, with a decrease in the number of ECM performed (around 30%). For this reason, we recommend performing this technique selectively, according to results of CT and PET scan.

Regarding the sensitivity of this procedure, when ECM is performed selectively, the sensitivity increases (from 0.44 to 0.76). This could probably be explained by the increase in the prevalence of N2–N3 disease in patients with enlarged lymph nodes or increased uptake in PET scan. Therefore, ECM sensitivity in the selective protocol is similar to other published series (ranging between 0.62 and 0.83) in which ECM was indicated in patients with enlarged lymph nodes [6–9].

Complications related to ECM are infrequent in all published reports [6–9]. In this updated series, with 65 additional patients, a low rate of complications persists (2.3%).

In conclusion, ECM is a useful and safe technique to assess nodal disease or direct tumour involvement in the subaortic or para-aortic stations. Its selective use according to the findings of CT and PET scans saves around 30% ECM, without decreasing the negative predictive value and accuracy.

ACKNOWLEDGEMENT

We thank Salvador Quintana, MD, from the Intensive Care Unit and Statistical Department of Mutua Terrassa University Hospital for his assistance in the statistical revision of this manuscript.

Conflict of interest: none declared.

REFERENCES

Author notes