What is the key question?

• Guidelines state that for patients with small (<3 cms), peripheral tumors without hilar involvement and a normal mediastinum on PET/CT scan; it is justifiable to proceed directly to surgical resection. However there is a paucity of data regarding the incidence of radiographic occult mediastinal disease detectable by EBUS-TBNA.
• The authors performed a metanalysis to evaluate the role of EBUS-TBNA in radiographically normal mediastinum and a total of 13 studies met inclusion criteria.
  
• The prevalence of radiographic occult mediastinal disease was 12.8% 95% CI, 10.4%-15.7%); sensitivity, 49.5% (95% confidence interval [CI], 36.4%-62.6%); NPV, 93.0% (95% CI, 90.3%-95.0%); and log DOR, 5.069 (95% CI, 4.212-5.925).
  
• Sensitivity improved for prospective multicentre studies but type of anaesthesia, imaging utilized to define negative mediastinum, and ROSE usage did not show association with sensitivity by regression analysis.
  
• Unfortunately, information was not consistently available to assess influence size, location and histology on prevalence or sensitivity.
  
• If EBUS-TBNA is negative the authors calculate a 7% probability of having occult mediastinal metastases. This is a marginal improvement from a false negative rate of 10% for PET/CT in systematic reviews, although of course this is not a direct comparison.

What is the bottom line?

• The authors performed a metanalysis to evaluate the role of EBUS-TBNA in radiographically normal mediastinum and a total of 13 studies met inclusion criteria.
  
• The prevalence of radiographic occult mediastinal disease was 12.8% 95% CI, 10.4%-15.7%); sensitivity, 49.5% (95% confidence interval [CI], 36.4%-62.6%); NPV, 93.0% (95% CI, 90.3%-95.0%); and log DOR, 5.069 (95% CI, 4.212-5.925).
  
• Sensitivity improved for prospective multicentre studies but type of anaesthesia, imaging utilized to define negative mediastinum, and ROSE usage did not show association with sensitivity by regression analysis.
  
• Unfortunately, information was not consistently available to assess influence size, location and histology on prevalence or sensitivity.
  
• If EBUS-TBNA is negative the authors calculate a 7% probability of having occult mediastinal metastases. This is a marginal improvement from a false negative rate of 10% for PET/CT in systematic reviews, although of course this is not a direct comparison.

Why read on?

• The requirement for EBUS-TBNA in patients with a radiographically normal mediastinum and early stage lung cancer in health care systems with limited access to EBUS-TBNA is a significant issue.
  
• The finding of a prevalence of radiographic occult mediastinal disease or 12.8% cannot be ignored and is explained by suboptimal sensitivity and specificity of PET/CT in staging the mediastinum.
  
• Although PET/CT data have primarily influenced guidelines to state that PET/CT underperforms in patients with larger tumors, central tumors or N1 disease, the performance of EBUS-TBNA to detect this radiographic occult disease in these situations needs to be explored further.