Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

Should all patients with non-small cell lung cancer who are surgical candidates have cervical mediastinoscopy preoperatively?

Three Part Question

In [patients undergoing lung resection for non-small cell lung cancer], would [routine cervical mediastinoscopy] reduce the incidence of [unnecessary thoracotomy].

Clinical Scenario

You are in a multidisciplinary team meeting, discussing a 76 year old lifelong smoker who has a T2 right upper lobe adenocarcinoma. She has COPD and arthritis and is quite a frail lady and lung function testing showed that she would not tolerate a pneumonectomy. The CT scan shows a 5cm tumour that may be resectable by lobectomy and there are no obviously enlarged mediastinal nodes although the radiologist reports that there are a few nodes there that are 0.8cm in diameter. A consultant surgeon accepts her for lobectomy, but the anaesthetist suggests a mediastinoscopy first to reduce the likelihood of an 'open and close' thoracotomy. The chest physicians state that this would be contrary to current guidelines and thus you suggest that you could look up the evidence and present it at the next week's meeting.

Search Strategy

Medline 1966-Oct 2005 using the Ovid interface
[exp mediastinoscopy/ OR] AND [exp neoplasm staging/ OR] AND [exp carcinoma,non-small-cell lung/ OR non small cell lung] limit to humans

Search Outcome

A total of 241 papers were found from the above search. We selected 4 systematic reviews and meta-analyses that summarized 14 diagnostic cohort studies on this topic. We identified 3 additional studies not included in the reviews and also summarized the largest cohort study in the literature. These are presented in the table

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Toloza et al,
Systematic review of Medline, Healthstar and Cochrane databases from 1991 to July 2001 of English language studies. Also handsearching performed. Searched for the performance of Transbronchial needle aspiration (TBNA), Transthoracic needle aspiration(TTNA), Endoscopic ultrasound-guided needle aspiration(EUS-NA) and mediastinoscopy. Compared method to either histology or long term follow up. Identified 5867 patients from 14 studies.Systemic Review and meta-analysis (Level 1a)Accuracy of Cervical Mediastinoscopy (14 studies)Sensitivity 0.81 (95%CI 0.76 to 0.85). Negative Predictive Value 0.91 (95%CI 0.58 to 0.97)Prevalence of positive nodal disease among papers varied from 30% to 88% No information about the types of patient included in each study, i.e. size of nodes identified on CT in the studies. Those patients undergoing TBNA or TTNA would have certainly had large identifiable nodes on CTs
Accuracy of trans-bronchial needle aspiration (12 studies )Sensitivity 0.76 (95%CI 0.72 to 0.79). Specificity 0.96 (95%CI 0.91 to 1.0)
Accuracy of Trans-thoracic needle aspiration (5 studies)Sensitivity 0.91 (95%CI 0.74 to 0.97). Negative Predictive Value 0.78 (95%CI 0.42 to 1.0)
Endoscopic ultrasound Needle aspiration (5 studies)Sensitivity 0.88 (95%CI 0.82 to 0.93). Specificity 0.91 (95%CI 0.77 to 0.97)
Comprehensive Systematic review into the Diagnosis and treatment of lung cancerSystematic review (Level 1a)Accuracy of CT staging of the mediastinumSensitivity 57% (95% CI 49-66%), Specificity 82% (95% CI 77-86%)A very well conducted systematic review with recommendations based only on the evidence presented.
Accuracy of staging by mediastinoscopy from 14 studiesSensitivity 81% (95% CI, 0.76-0.85). Specificity 100% (unreliable). Negative predictive value 91% (range, 58-97%) negative (FN) rate 10%.
RecommendationsCT enables the detection of enlarged mediastinal nodes, but the poor specificity makes tissue sampling necessary to determine the patients true nodal status if surgery is a therapeutic option (Level Ib, II and III DS).
Cost-effectivenessMediastinoscopy is more effective but also more costly than selecting patients for mediastinoscopy on the basis of their CT results. Routine mediastinoscopy may not be cost-effective in T1 patients even if it is cost-effective for T2 and T3 patients
Performance of PETSensitivity 84% (CI 0.78-0.89), Specificity 89% (CI 0.83-0.93) PPV 79% and NPV93%
Comprehensive systematic review, updating previous guidelines by SIGN from 1998Systematic review (level 1a)CT for the assessment of the mediastinumFor all categories of pts with NSCLC the reliability of CT is poor…with false negative rate of 13%
RecommendationPatients with small peripheral tumours and a negative CT scan of the mediastinum require no further investigation. Otherwise it is reasonable to further investigate the mediastinum with mediastinoscopy
Detterbeck et al,
Guideline for the Multidisciplinary Thoracic Oncology Program, based on the evidence provided by Toloza et alSystematic review (level 1a)In patients suspected of having NSCLC, who have normal mediastinal nodes by CT, but in whom invasive staging of the mediastinum is recommended (because of a high FN rate of CT), mediastinoscopy is the invasive procedure of choice to rule out mediastinal node involvement.This recommendation is based on the ability of mediastinoscopy to stage most of the commonly involved mediastinal node stations with a low FN rate (approximately 10%) and low morbidity (2%; outpatient procedure). Level of evidence, fair; benefit, substantial; grade of recommendation, B
De Leyn et al,
235 patients with operable NSCLC and no enlarged mediastinal lymph nodes (i.e. >15mm) on CT scanning. (Radiological N0 patients) 355 patients assessed by CT scanning with 8mm slices, and 5mm at the hilum and contrast Patients with T1N0 or T2N0 squamous cell carcinoma excluded.Diagnostic Cohort study (Level 2b)Cervical mediastinoscopy resultsPositive histology in 47/235 patients (20%). 37 patients has extranodal or multilevel N2 positive diseaseAll but 19 Patients with T1N0 or T2N0 squamous cell carcinoma were excluded.
Results by pre-operative stagingCervical mediastinoscopy was positive in : 9.5% of cT1N0 cases, 17.7% in cT2N0 cases 31.2% of cT3N0 cases 33.3% of cT4N0 cases
Final pathological staging20 patients (11%) with negative CT and Mediastinoscopy had intra-operative diagnosis of N2 disease.
Choi et al,
South Korea
291 patients with stage 1 NSCLC (T1-2, N0) after clinical evaluation and CT examination (Absence of lymph nodes >1mm in shortest diameter) Mediastinoscopy of stations 2L, 2R, 4L, 4R, and subcarinal stations and frozen section, followed by thoracotomy at same session.Retrospective Cohort study (level 2b)Positive Mediastin-oscopy20/291 (6.9%) of patients had N2 or N3 disease after mediastinoscopyFrozen section rather than formal histology was used to assess mediastinoscopy samples
Negative Mediastin-oscopy25/271 (9.2%) of patients had N2 disease on thoracotomy.
Complications5 patients had hoarseness, 1 wound infection.
Daniels et al,
76 patients with NSCLC from 1997 to 1999, with Negative Mediastinal CT scan. All patients had routine Mediastinoscopy except for 10 patients with T1 Squamour Cell Ca. Intraoperatively full mediastinal lymph node dissection was performedRetrospective Cohort study (level 3b)Positive Mediastinoscopy16 of 66 pts had positive mediastinoscopy (24%)
Negative Mediastinoscopy3 pts with negative mediastinoscopy had positive subcarinal nodes (station 7)
Hammound et al,
Retrospective review of 1745 mediastinoscopies for lung cancer from 1988 to 1998.Retrospective Cohort Study (level 2b)Patients with negative mediastinoscopy76/947 (8.0%) pts had N2 disease intraoperatively9 patients had negative frozen section mediastin-oscopy, but histologically malignant disease was found
Complications of 2137 mediastinoscopiesDeath 1/2137 ( 0.05%) directly due to mediastinoscopy.12 complications (0.5%) including 6 arrhythmias, 1 pneumothorax, 2 bleeding, 1 oesophageal perforation.


Accurate nodal staging, particularly N2 status is of paramount importance in the selection of patients for surgery. Studies have reported that around 20% of patients undergoing clinical staging are either overstaged or understaged when pathological staging is compared, and 20% of preoperative N0 patients may actually have pathological N1 or N2 status (Toloza). Thus an optimal strategy for mediastinal staging is vital. The most important paper identified was the NICE guidance on lung cancer management and treatment, published in February 2005. They used the results from their own systematic review and that published by Toloza together with a small number of additional studies to calculate the sensitivity and specificity for either CT scanning or mediastinoscopy in the diagnosis of nodal status prior to thoracotomy. The sensitivity of CT scanning across 20 studies was 57% and the specificity was 82%. Mediastinoscopy across 14 studies showed a sensitivity of 81% and a false positive rate of less than 10% (The specificity is by definition 100% as the gold standard of positive histology is obtained by mediastinoscopy). They also performed a cost-effectiveness analysis of routine mediastinoscopy and found it to be cost effective for T2 and T3 tumours. NICE however concludes that mediastinoscopy should only be performed for patients with mediastinal lymph nodes >1cm in the shortest axis identified on CT scanning. Finally they also report that a negative PET scan does not need tissue confirmation but a positive PET scan should undergo confirmation with mediastinoscopy. The Scottish Intercollegiate Guidelines Network also published an update of their guidelines in 2005. They collaborated with NICE but were more liberal in their recommendation for mediastinoscopy. They state that due to the high false negative rate of CT scanning, 'Patients with small peripheral tumours and a negative CT scan of the mediastinum require no further investigation. Otherwise it is reasonable to further investigate the mediastinum with mediastinoscopy, Grade B.' Based on the Systematic review by Toloza et al, Detterbeck et al published a guideline for the American College of Chest Physicians. They state that for patients with a negative mediastinal CT, mediastinoscopy is the invasive procedure of choice to assess the mediastinal nodes due to its low false negative rate and morbidity (Grade B). The ACCP also recommends that good practice dictates lymph node samples from high and low paratracheal nodes (2L,2R,4L,4R) pretracheal nodes (stations 1&3) and the anterior subcarinal nodes (station 7) when performing mediastinoscopy. De Leyn et al reported the findings of mediastinoscopy in patients with adenocarcinoma with a negative CT scan (no lymph nodes over 15mm in long axis). Mediastinoscopy diagnosed 47 patients (20%) with N2 disease, thus avoiding unnecessary surgery. 10% of patients with T1 disease had mediastinoscopic N2 disease. Also of concern is the fact that despite routine CT and mediastinoscopy another 20 patients (11%) had an intraoperative diagnosis of N2 disease. The Number Needed to Treat with mediastinoscopy to prevent an unnecessary operation was 5 in this study. Choi et al performed routine mediastinoscopy and frozen section prior to resection in 291 patients with stage I disease after negative mediastinal CT scanning. 20 patients had positive frozen section (7%) and had their operation cancelled. Interestingly an additional 25 patients (9%) had N2 disease after intra-operative assessment and formal histology. Daniels et al instituted a protocol of routine mediastinoscopy for all patients with negative mediastinal CT. 16 out of 66 (24%) of patients had positive mediastinoscopy, detecting N2 disease in patients previously categorized as N0. The largest review of mediastinoscopy was by Hammound et al.They reported their experience with 2137 patients. There was a single death due to the procedure (0.05%) and 12 complications (0.5%) which included 6 arrhythmias one oesophageal perforation and 2 bleeds, thus demonstrating that mediastinoscopy is a relatively safe procedure. Interestingly they also had 9 patients where the mediastinoscopic frozen section was negative but formal histology proved to be positive. Thus in summary CT scanning has a high false negative rate. NICE recommends routine mediastinoscopy for patients with nodes >1mm on CT scanning but SIGN recommends mediastinoscopy in all but small peripheral tumours, and the ACCP recommend mediastinoscopy for all patients considered for surgery with a negative mediastinal CT scan if PET scanning cannot be performed.

Clinical Bottom Line

Patients with resectable non-small cell lung cancer who have had a negative mediastinal CT scan should all undergo mediastinoscopy. The Number Needed to Treat with mediastinoscopy to prevent an unnecessary thoracotomy is around 5-15 patients. Exceptions to this may be patients with a T1 tumour, patients with a small peripheral tumour or patients who have had a negative PET scan.


  1. Toloza EM, Harpole L, Detterbeck F, McCrory DC. Invasive staging of non-small cell lung cancer: a review of the current evidence. Chest 2003;123(1 Suppl):157S-166S.
  2. Commissioned by the National Institute of Clinical Excellence. The diagnosis and treatment of lung cancer : Methods, Evidence and guidance 2005. Published by the National Collaborating Centre for Acute Care at The Royal College of Surgeons of England, 35-43 Lincoln's Inn Fields, London, WC2A 3PE.
  3. Scottish Intercollegiate Guidelines Network. Management of patients with lung cancer : A national clinical guideline,2005. Scottish Intercollegiate Guidelines Network Royal College of Physicians, 9 Queen Street Edinburgh EH2 1JQ.
  4. Detterbeck FC, DeCamp MM, Jr., Kohman LJ, Silvestri GA. American College of Chest Physicians Lung cancer. Invasive staging: the guidelines. Chest 2003;123(1 Suppl):167S-175S.
  5. De Leyn P, Vansteenkiste J, Cuypers P, Deneffe G, Van Raemdonck D, Coosemans W, Verschakelen J, Lerut T. Role of cervical mediastinoscopy in staging of non-small cell lung cancer without enlarged mediastinal lymph nodes on CT scan. European Journal of Cardio-Thoracic Surgery 1997;12(5):706-12.
  6. Choi YS, Shim YM, Kim J, Kim K. Mediastinoscopy in patients with clinical stage I non-small cell lung cancer.[see comment]. Annals of Thoracic Surgery 2003;75(2):364-6.
  7. Daniels JM, Rijna H, Postmus PE, van Mourik JC. Mediastinoscopy as a standardised procedure for mediastinal lymph node staging in non-small cell lung carcinoma. European Journal of Cardio-Thoracic Surgery 2001;19(3):377-8.
  8. Hammound ZT, Anderson RC, Meyers BF, Guthrie TJ, Roper CL, Cooper JD, Patterson GA. The current role of mediastinoscopy in the evaluation of thoracic disease. Journal of Thoracic & Cardiovascular Surgery 1999;118:894-899.