Lung cancer is one of the most prevalent malignancy worldwide with significant cancer related mortality especially in the western world.¹ Non-small cell lung cancer (NSCLC) accounts for the majority of patients with the remainder as mostly small cell lung cancer (SCLC). Small cell lung cancer variant is usually metastatic at presentation. This variant is chemo responsive and is managed with chemotherapy and radiation therapy; thus it does not require staging for management. NSCLC, however, is managed with surgery, neoadjuvant chemoradiation or palliative care as per the disease stage and thus correct mediastinal staging is vital for treatment planning.²

The survival rate in NSCLC varies from > 70% in localized disease where surgical resection is possible to < 30% in stage III (locally advanced disease). Thus, accurate diagnosis and staging is of paramount importance for both prognostic and therapeutic reasons.²

Mediastinal staging conventionally relied heavily on invasive modalities like mediastinoscopies, and thoracotomies. Endoscopic ultrasound (EUS) evaluation of mediastinum, EUS guided fine needle aspiration (EUS-FNA) and endobronchial ultrasound guided transbronchial needle aspirations (EBUS-TBNA) have evolved over last few years as novel and minimally invasive modalities for accurately staging mediastinal nodes, to guide appropriate therapy, and to avoid unnecessary surgeries in NSCLC patients.³

According to 7th edition of TNM (Tumor, Nodal status, Metastasis) staging for NSCLC, stage I and II patients are treated with surgical resection, whereas stage III (N2 nodal status, T4 mediastinal invasion) are offered chemoradiation with only limited role of surgical resection.² Mediastinal nodal sampling by using EUS/EBUS has been documented to be superior to surgical staging in published literature.³ This has been emphasized also in the latest 2013 guidelines for lung cancer by American College of Chest Physicians which state that EUS/EBUS are the techniques of choice for mediastinal staging.^4,5

NSCLC is staged according to the tumor-node-metastasis (TNM) system currently in its 7th edition (Table 1). This system takes into account the characteristics of the local tumor (T), the presence or absence of regional lymph node metastasis (N), and the presence or absence of distant metastases (M). The stage of the tumor (stage I through IV) depends upon the particular combination of T, N, and M characteristics for the given patient.

EUS can contribute to each component of TNM staging for lung cancer. It can help characterize the primary tumor (in centrally located tumors), assesses the mediastinal lymph nodes for evidence of metastatic disease, and evaluates some sites of distant metastasis such as the left lobe of the liver and adrenal glands. Among these contributions, however, mediastinal lymph node evaluation is its primary role.

EUS aids in biopsy of intrapulmonary tumors in tumors located centrally near or adjacent to esophagus. In a prospective study of 32 patients with suspected lung cancer and a primary tumor located adjacent to the esophagus, intrapulmonary masses were detected in all patients, and the diagnosis of lung cancer was established in 97% of patients.⁶ In another small retrospective study, EUS identified and diagnosed all periesophageal pulmonary masses.⁷

Once the primary tumor has been identified, EUS can help to define mediastinal invasion, which includes involvement of mediastinal structures such as left atrium, large central vessels, esophagus, and vertebrae by the intrapulmonary tumor. This invasion if present places the patient in T4 category (stage IIIb) and generally precludes surgical resection as a treatment option.⁸

EUS has a sensitivity of 87% and specificity of 98% to detect T4 mediastinal invasion in current literature.⁹ This is significantly high when compared to a preoperative computed tomography (CT) scan, which has a low sensitivity (< 75%) to detect mediastinal invasion and positron emission tomography (PET) scan which does not have a defined role in T4 staging because of poor anatomic resolution.^10,11

Mediastinal/hilar nodal involvement (N stage) by the tumor is an important determinant for staging and guiding treatment. Lymph node sampling for histopathological examination is necessary in patients with enlarged mediastinal lymph nodes on CT scan or metabolically active nodes on PET scan (Fig. 1), as imaging modalities alone have a low accuracy in staging of mediastinal nodes.

CT traditionally has been the first modality to identify and define mediastinal and pulmonary lesions. However, it relies mainly on morphological characteristics and nodal size (> 10 mm in short axis) to define abnormal lymph nodes. This results in a low accuracy because of high false positive and false negative rates of approximately 25%.^12?14

In two systematic analyses, cumulative sensitivity of CT scan to detect mediastinal nodal involvement was 55%?61% and specificity of 79%?81%.^8,15

Integration of PET-CT has improved these values to some extent. In a recent meta-analysis, the pooled weighted sensitivity and specificity of PET-CT in a patient-based group were estimated 76% and 88%, respectively.¹⁶

PET-CT, however, has not been able to solve the problem of using nodal size as the defining criterion for pathological nodes resulting in high false positive rates especially in large size lymph nodes. In one study the sensitivity of PET-CT was significantly higher among enlarged (> 10 mm) than non-enlarged (≤ 10 mm) lymph nodes (74% vs 40%). On the other hand, specificity (81% vs 98%) and accuracy (78% vs 90%) were significantly lower in enlarged compared to non-enlarged lymph nodes.¹⁷ Another study showed that upto 16% patients had occult N2 disease, which was previously staged N2/N3 negative by PET-CT.¹⁸ Thus PET-CT alone cannot replace pathological sampling of lymph nodes by EUS-FNA/surgery. In selected patients who are N2 negative by PET-CT, as well as patients with enlarged positive N2 nodes, sampling of mediastinal nodes is still necessary before a final diagnosis is made.

EUS-FNA is effective at detecting and staging mediastinal metastatic disease. It can sample lymph nodes in the posterior mediastinum (level 4L, lower left paratracheal; level 6, para aortal; level 8, para esophageal; and level 9, near inferior pulmonary ligament) and subcarina (level 7), sites that are particularly susceptible to metastasis (Fig. 1, 2). In addition, it might be able to sample lymph nodes in the aortopulmonary window (level 5), although this is challenging in a few cases because of interposition of pulmonary artery, which makes sampling technically difficult.

EUS visualization is limited in superior and anterior mediastinum, especially upper paratracheal (level 2) and lower paratracheal nodes to the right (level 4R) due to interposition of air filled bronchi. This precludes sampling from these stations using EUS alone, and a combined approach using EUS + EBUS is a preferred modality in such situations. In addition to lymph nodes, EUS can be used to sample left adrenal gland, left liver lobe metastasis and also centrally located intrapulmonary tumors as discussed earlier.

The role of EUS in patients with NSCLC and a negative CT finding for enlarged mediastinal nodes is still not clear, as most data for EUS-FNA is in patients with enlarged mediastinal lymph nodes. However, some emerging data have shown importance of EUS evaluation in these patients, as approximately 20% of these normal size nodes can be positive for malignancy.^18?20 In a retrospective cohort study of 69 patients with NSCLC and negative mediastinum on CT, EUS detected advanced disease, including N2/N3 disease, T4 disease and distant metastasis in almost 25% of patients.²¹ In another prospective study of 76 patients with negative CT scan, EUS evaluation impacted management of 25% of patients, and precluded surgery in 12% of patients.²²

Accuracy of mediastinal lymph nodal staging by EUS-FNA in lung cancer has been extensively investigated in a number of studies.^23?33 In a meta-analysis of 18 studies of EUS-FNA for the mediastinal staging of lung cancer, overall sensitivity was 83% and specificity was 97%.²³ The overall accuracy of EUS-FNA (i.e., weighted mean of sensitivity and specificity) is of 83% to 97% and a sensitivity of 84% to 92%. This was illustrated by a prospective cohort study of 104 patients in which EUS-FNA detected malignant posterior mediastinal lymph nodes with a sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of 92%, 100%, 100%, 94%, and 97%, respectively.²⁸

Majority of literature concerned the patients with enlarged (> 1 cm) mediastinal lymph nodes at CT/PET-CT. Few studies have focused specifically on small (short axis ≤ 10 mm) nodes; sensitivity in this group has varied between 35% and 93%.^21?23,34 The pooled sensitivity in a meta-analysis for this subgroup was 58% (95% confidence interval, 39% to 75%).²³

Lung cancer patients can commonly (~40%) present with distant metastasis to brain, bone, adrenal glands, and liver. EUS is an effective modality to screen and sample metastasis from celiac group of nodes, left adrenal gland and left lobe of liver.

In one prospective cohort study of 113 patients with lung cancer, celiac lymphadenopathy was identified in 19 patients (11%).³⁵ The diagnostic yield of EUS-FNA for detecting metastases to these lymph nodes was 100%, compared to only 50% by CT scan.

In a retrospective study, EUS-FNA guided sampling of pleural effusion showed positive malignant cytology in 7 out of 10 patients with NSCLC.³⁶ EUS-FNA thus, has clearly demonstrated its utility as a minimally invasive modality to correctly identify stage 4 NSCLC.

In our own data of 21 patients with left adrenal enlargement, 6 patients were found to have metastatic lung cancer deposits on adrenal sampling.³⁷ In another study of 40 patients suspected with lung cancer and an enlarged left adrenal gland, EUS-FNA altered TNM staging in 70% of patients.³⁸

Detection of liver, celiac and adrenal deposits on EUS defines M1 stage of the disease and excludes curative surgery. EUS thus is a unique modality wherein abdominal evaluation for such lesions can be done simultaneously during a mediastinal staging procedure.

EUS can help in restaging of disease in patients with stage III disease after neoadjuvant therapy. Some of these patients may become eligible for surgical resection after chemoradiation. In a prospective study, residual NSCLC was detected by EUS-FNA in mediastinal nodes with sensitivity and specificity of 75% and 100% respectively.³⁹ EUS-FNA is superior to PET for restaging mediastinal metastasis as it confirms rather than exclude metastasis.^40,41

Both EUS and EBUS are complementary to each other in mediastinal evaluation of lung cancer patients. Combined together both techniques can virtually reach almost all nodal stations of mediastinum (Fig. 2⁴¹, Table 2^42,43). In general EUS is an excellent modality for visualization and sampling from posterior and inferior mediastinum whereas EBUS is a preferred modality in anterior mediastinum.

An interesting emerging concept is to use a single EBUS scope to image and sample nodes through bronchoscopy first and subsequently use the same scope through esophagus as EUS scope. This concept is financially and logistically lucrative and would enable a single operator to do a complete medical mediastinoscopy. The emerging data have been promising with a sensitivity of about 90% in staging of NSCLC.^44?46

EUS impacts management of approximately 95% of patients with lung cancer, and has a major role in preventing unnecessary mediastinoscopies and futile thoracotomies.^26?28,47,48 In a prospective study of 84 patients with mediastinal masses suspected of being malignant, EUS prevented thoracotomy or thoracoscopy in 48% and mediastinoscopy in 68% of patients by demonstrating lymph node metastases.²⁶ In another study of 59 patients, all scheduled for mediastinoscopy, EUS-FNA proved mediastinal metastases in 39% of patients, and mediastinoscopy was eventually performed in only 22%.⁴⁸

In another prospective study of 242 patients, EUS-FNA demonstrated lymph node metastases, tumor invasion, or an alternative diagnosis in 70% of patients, thus preventing further surgical interventions.⁴⁹

• In patients with high suspicion of N2, N3 involvement, either by discrete mediastinal lymph node enlargement or PET uptake (and no distant metastases), EBUS-FNA, EUS-FNA or combined EBUS/EUS-FNA is recommended over surgical staging as a mostly suitable diagnostic modality (grade 1B).⁵

• In patients with an intermediate suspicion of N2, N3 involvement, i.e., a radiographically normal mediastinum (by CT and PET) and a central tumor or N1 lymph node enlargement (and no distant metastases), EBUS-TBNA, EUS-FNA or combined EBUS/EUS-FNA is suggested over surgical staging as a mostly suitable diagnostic modality (grade 2B).⁵

Molecular analysis and targeted therapy for different subtypes of NSCLC are emerging areas with lot of potential for therapeutic application. Samples obtained from mediastinal lymph nodes by EUS-FNA can be used to detect lung cancer-associated genes, such as carcinoembryonic antigen, cytokeratin 19, KS1/4, lunx, muc 1, and prostate derived ETS factor. In one study, overexpression of the KS1/4 gene was associated with metastatic disease, even in cytology-negative lymph nodes.⁵⁰ Identifying the presence of somatic epidermal growth factor receptor (EGFR) and kras mutations is important for predicting a therapeutic response to the EGFR tyrosine kinase inhibitors in patients with lung cancer. EUS-FNA samples have been successfully used to test for the presence of these mutations.⁵⁰

Making combined EUS/EBUS more accessible and acceptable to both gastroenterologists and pulmonologists with focused training programs to learn these techniques is a challenge, which needs to be overcome to achieve success against this deadly cancer.