At one time, specialists were separated by clear divisions of the human anatomy. Gastroenterologists, like their general surgery colleagues, worked only below the diaphragm. With the introduction of flexible endoscopy, however, it soon became apparent that gastroenterologists rather than their pulmonary colleagues should treat diseases of the esophagus, since they obviously needed to pass this organ to access the stomach. Now, the gastroenterologist's area of practice is set to expand again because endoscopic ultrasound (EUS) gives gastroenterologists much better access to portions of the mediastinum than their surgical or pulmonary colleagues can achieve. This is a natural evolution in the development of EUS applications. The procedure has traditionally been used to stage esophageal cancers and has been fairly successful in this area since it can stage periesophageal and mediastinal lymph nodes.
In most situations, EUS can be used to stage important pulmonary lymph nodes that are found in paratracheal stations 2 and 3, aortopulmonary window station 4L, subcarinal station 7, periesophageal station 8, and inferior pulmonary ligament station 9. EUS cannot stage lymph node stations more distally in the pulmonary tree, such as stations 10 through 15.
To access all of the lymph node stations needed to stage lung cancers, it is necessary to use a combination of EUS with fine-needle aspiration (FNA) along with endobronchial ultrasound (EBUS). The endobronchial route allows access to most of the important stations, with the exception of stations 5 and 6. That said, EBUS is currently limited in the performance of FNA: It must be performed blind rather than under real-time guidance (although the latter is being developed). This combination of procedures could most likely substitute for mediastinoscopy. However, it has not been assessed in a prospective fashion whether EUS and EBUS can perform as well as the more invasive procedure. Moreover, mediastinoscopy can access station 5 more readily than either of these two techniques.
The meta-analysis done by Ogita et al assesses the sensitivity, specificity, positive, and negative predictive value of EUS with FNA in multiple studies staging non-small-cell lung cancers. This review demonstrates the consistent superiority of EUS with FNA as compared with mediastinoscopy, computed tomography (CT), and positron-emission tomography (PET) as a staging procedure. The information is useful, allowing the investigator to compare multiple small studies to give a better overall assessment of the utility of these modalities.
Meta-analysis also has known limitations, as it is very dependent on the studies included. The first important limitation is that these studies are all being conducted at expert centers. It has been shown in a number of prior EUS studies that EUS often performs worse in actual clinical practice than in expert hands. Training, interest in the technology, patient selection, and resources such as onsite cytology interpretation are all factors known to affect test results.
Secondly, the performance of EUS with FNA in relationship to CT and PET really depends on what is found by those studies. For instance, another recent meta-analysis found that the accuracy of PET is dependent on the size of lymph nodes found on CT. Even if PET is negative, if the nodes are > 16 mm in size, they are likely to contain cancer in 21% of the patients. On the other hand, if PET is negative and the nodes are 10 to 15 mm in diameter, the likelihood of cancer is 5%. This leads to changes in clinical evaluation: If the cancer is peripheral, PET is negative, and nodes are < 15 mm, further mediastinal evaluation is unlikely to be of value. Most of the evaluation strategies included in this review were done using EUS and FNA for evaluation of suspicious nodes, although the criteria for suspicious nodes varied a great deal. In fact, one of the studies used by the authors only studied patients with negative CT and PET.
Obviously, sensitivity and specificity varies with the patient population studied. Lung tumor location also influences likely location of nodal spread. For instance, if the tumor is located in the left upper lobe, metastasis to the level 6 para-aortic or level 5 aortopulmonary window is likely. Thus, in these EUS with FNA studies, knowing the location of the primary tumor could influence the decision to proceed with EUS. If metastasis to an unapproachable area were likely, these patients may well have been excluded or a bias may exist regarding their inclusion. These hidden biases can be determined with the use of a propensity test, to be certain that the patients undergoing EUS were no different in terms of disease stage or location than those who did not undergo such investigation.
Dr. Wang receives research support from Olympus.