What Are Mediastinal Lymph Nodes and Why Do They Swell?

Mediastinal lymph nodes are clusters of immune tissue scattered throughout the mediastinum, the central compartment of the chest between the lungs. They filter lymphatic fluid draining from the lungs, airways, esophagus, and heart, and they are among the first places clinicians look when staging lung cancer, evaluating unexplained chest findings, or working up systemic diseases like sarcoidosis and lymphoma. Whether your doctor mentioned them after a CT scan or you saw the term on an imaging report, understanding what these nodes do, what makes them swell, and how they’re investigated fills in a picture that radiology shorthand often leaves frustratingly vague.

Location and Organization

The mediastinum is essentially everything in your chest that is not lung tissue. It houses the heart, the major blood vessels, the trachea, the esophagus, and a generous network of lymph nodes. These nodes are not randomly scattered; they follow the airways and blood vessels in a pattern that surgeons and radiologists map using a standardized numbering system. The current standard is the International Association for the Study of Lung Cancer (IASLC) lymph node map, which assigns each node station a number from 1 through 14 and groups those stations into broader “zones.”¹ That map replaced earlier, regionally different systems so that a surgeon in Tokyo and a radiologist in Chicago describe the same node station the same way.²

In broad strokes, the upper mediastinal stations (1 through 4) sit along the trachea and above the aortic arch. The aortopulmonary stations (5 and 6) flank the aorta and pulmonary artery. The subcarinal station (7) sits just below where the trachea splits into the two main bronchi. Stations 8 and 9 run along the esophagus and the pulmonary ligament, lower in the chest. Stations 10 through 14 are technically hilar and intrapulmonary nodes, sitting within the lung tissue itself. The numbering matters because cancer that has reached station 7 has different treatment implications than cancer that has only reached station 10, even though the two are only centimeters apart.

Lymphatic Drainage and Why It Is Not Always Predictable

You might assume that fluid and cells from the lung drain through a tidy chain: first through intrapulmonary nodes, then hilar nodes, then mediastinal nodes. That is the usual sequence, but anatomy doesn’t always cooperate. A cadaver study of 260 adults found that roughly a quarter of lung segments had direct lymphatic channels running straight to mediastinal nodes, bypassing the hilar stations entirely.³ These shortcut pathways were more common in the upper lobes. This anatomic quirk has real clinical consequences: it helps explain why lung cancers sometimes show up in mediastinal nodes without first appearing in the closer hilar nodes, a phenomenon called “skip metastasis.”

What Counts as a Normal Size

On a CT scan, mediastinal lymph nodes are measured by their short-axis diameter in the transverse (cross-sectional) plane. The widely cited upper limit of normal is 10 mm. That threshold comes from studies mapping normal node dimensions in patients without known malignancy, where nodes above 10 mm on the short axis stood out from the rest of the size distribution.⁴ A separate anatomic and CT study found that virtually all normal nodes measured under 16 mm in their largest dimension, with the average cadaveric node measuring about 13 by 8 mm.⁵

The 10 mm cutoff is a guideline, not a verdict. A node measuring 12 mm could be perfectly reactive from a recent cold, while a 9 mm node could harbor microscopic cancer. Size alone does not distinguish benign from malignant. That ambiguity is the central headache of mediastinal lymph node assessment and the reason imaging is often just the first step rather than the last word.

Benign Causes of Enlargement

Enlarged mediastinal lymph nodes are common on chest CT scans, and most of the time they are not cancer. Reactive enlargement from ordinary respiratory infections is probably the most frequent cause. Beyond that, several specific conditions have well-recognized patterns of mediastinal lymph node involvement.

Sarcoidosis is the classic example. It produces non-caseating granulomas and has a striking tendency to involve certain node stations: lower paratracheal nodes (station 4) were affected in over 97% of sarcoidosis patients in one imaging study, with hilar nodes (station 10) involved about 78% of the time and subcarinal nodes (station 7) in roughly 77%.⁶ Bilateral hilar lymph node enlargement is so characteristic of sarcoidosis that it often prompts the diagnosis even before a biopsy is done.

Fungal infections, particularly histoplasmosis, can also enlarge mediastinal nodes and mimic both sarcoidosis and malignancy on imaging. Histoplasmosis is endemic in certain river valleys and can present with mediastinal lymph node enlargement that looks worrisome enough to trigger a biopsy or even surgery before the infection is identified. Tuberculosis behaves similarly, favoring mediastinal and hilar nodes and occasionally producing calcified nodes over time.

Occupational dust exposures leave their own signature. Silicosis and coal-worker’s pneumoconiosis can produce a distinctive “eggshell calcification” pattern, where the outer rim of the node calcifies while the center stays soft. That peripheral calcification pattern is uncommon enough in other conditions that seeing it on a chest X-ray or CT usually points straight to a dust-exposure history.⁷

In rare cases, chronically inflamed mediastinal nodes can rupture their contents into the surrounding tissue and trigger fibrosing mediastinitis, a condition in which dense scar tissue gradually encases and compresses mediastinal structures like the superior vena cava, pulmonary veins, or airways.⁸ Histoplasmosis is the most frequently implicated trigger, though many cases are idiopathic.

Lung Cancer and Mediastinal Node Staging

The reason mediastinal lymph nodes get so much clinical attention is lung cancer. Whether cancer has reached the mediastinal nodes is one of the single biggest factors in determining treatment. A lung tumor confined to the lung with negative mediastinal nodes may be surgically curable. The same tumor with cancer in mediastinal nodes on the opposite side of the chest is generally treated with chemotherapy, radiation, or immunotherapy rather than surgery alone.

The TNM staging system classifies mediastinal node involvement as N2 (cancer in nodes on the same side as the primary tumor) or N3 (cancer in nodes on the opposite side, or in certain supraclavicular nodes). In one study of patients with suspected advanced disease, about two-thirds were confirmed to have N2 or N3 involvement after definitive staging.⁹

Skip Metastasis

As the anatomy discussion hinted, cancer does not always march through nodes in an orderly sequence. “Skip metastasis” refers to cancer appearing in mediastinal nodes (N2) while the hilar nodes (N1) remain clean. One large study found that roughly 17% of non-small cell lung cancer patients with mediastinal involvement had this pattern.¹⁰ An earlier, smaller series put the number higher, at about 33% of N2 patients, with upper mediastinal stations 1, 2, and 3 being the most common skip destinations and the subcarinal station 7 being particularly frequent in right-sided middle and lower lobe tumors.¹¹ The clinical significance is debated; the larger study found that skip metastasis was not by itself a prognostic factor, meaning these patients did not have clearly better or worse outcomes just because the hilar nodes were spared.

Lymphoma

Hodgkin lymphoma and non-Hodgkin lymphoma both involve mediastinal nodes but in characteristically different patterns. Hodgkin disease has a strong predilection for superior mediastinal nodes, including the prevascular, pretracheal, and aortopulmonary stations. In fact, if a patient with intrathoracic lymph node enlargement has no involvement of the superior mediastinum, the diagnosis of Hodgkin disease becomes less likely. Hodgkin disease also tends to spread contiguously, from one node group to the next in sequence.¹² Non-Hodgkin lymphoma is less predictable. It more commonly involves posterior mediastinal and paracardiac nodes, is more likely to involve just one node group, and its spread tends to skip node stations rather than moving through them one by one.¹³

Metastases from Cancers Outside the Chest

Mediastinal lymph nodes do not only collect trouble from the lungs. Cancers originating elsewhere in the body can metastasize to the mediastinum. Breast cancer is the most common extrathoracic source, followed by kidney and prostate cancers.¹⁴ When patients with a known cancer elsewhere develop enlarged mediastinal nodes, clinicians face a dilemma: the nodes could represent metastatic disease, but they could also be reactive or related to an entirely separate condition. In one study, only about a third of PET-positive mediastinal nodes in patients with extrathoracic cancers turned out to be malignant, with the remaining two-thirds being benign.¹⁵ That makes tissue confirmation important before changing a patient’s treatment plan based on imaging alone.

How Mediastinal Lymph Nodes Are Evaluated

Chest CT is the starting point. It tells you where nodes are, how large they are, whether they are calcified, and what pattern they form. But as we’ve seen, a node’s size on CT cannot reliably tell you whether it is malignant. PET/CT, which highlights tissues with high metabolic activity, adds another layer. Infections, sarcoidosis, and even post-surgical inflammation can all light up on PET, producing false positives. Conversely, slow-growing tumors and small metastatic deposits can fail to show activity, producing false negatives.¹⁶

A study focused on lung cancer staging found that PET/CT had a sensitivity of about 74% and specificity of about 73% for detecting mediastinal and hilar node metastases, with nearly half of PET-positive cases turning out to be false positives on surgical confirmation.¹⁷ Those numbers underscore why imaging alone rarely settles the question of whether a mediastinal node is malignant. When the answer matters for treatment, tissue sampling is usually necessary.

Getting a Tissue Diagnosis

The gold standard for decades was cervical mediastinoscopy, a surgical procedure in which a small scope is inserted through an incision above the breastbone to biopsy accessible mediastinal nodes. Mediastinoscopy provides larger tissue samples and can reach several upper and subcarinal stations effectively. However, it requires general anesthesia, an operating room, and a recovery period.

Endobronchial ultrasound-guided transbronchial needle aspiration, or EBUS-TBNA, has largely replaced mediastinoscopy as the first-line approach. It uses a bronchoscope equipped with an ultrasound probe, threaded through the airway, to visualize nodes through the bronchial wall and sample them with a fine needle. It can be performed under moderate sedation as an outpatient procedure. Systematic reviews report that EBUS-TBNA achieves sensitivity in the range of 85 to 94% with specificity approaching 100%, and the complication rate sits around 2%.¹⁸ Most current guidelines now recommend it as the first-line method for invasive mediastinal staging.¹⁹

In at least one head-to-head comparison, EBUS-TBNA achieved 87% sensitivity versus 68% for mediastinoscopy, though the difference in correctly determining the overall pathologic node stage was not statistically significant.²⁰ Mediastinoscopy still has a role when EBUS results are negative but clinical suspicion remains high, or when larger tissue samples are needed for molecular testing.

EBUS-TBNA has also proven valuable beyond lung cancer staging. In patients with extrathoracic malignancies and suspicious mediastinal nodes, the procedure diagnosed mediastinal metastases in about 44% of cases while ruling out malignancy or providing an alternative benign diagnosis in the rest.²¹

Sampling Versus Full Dissection During Lung Surgery

When a patient with lung cancer goes to the operating room for tumor removal, the surgeon has a choice about what to do with the mediastinal nodes. Mediastinal lymph node sampling means removing only nodes that look or feel suspicious. Mediastinal lymph node dissection means systematically removing all node-bearing tissue from defined stations, regardless of appearance. The debate over which approach is better has played out over multiple trials.

A multicenter study found that sampling alone identified only about half of N2 disease that full dissection would have caught. Full dissection found 60 patients with N2 disease in a group where sampling had flagged only 31.²² That argues strongly for dissection as a staging tool: you cannot plan the right post-surgical treatment if you do not know the full extent of node involvement.

On the other hand, a large randomized trial (the ACOSOG Z0030 trial) focused specifically on patients whose nodes had already been thoroughly sampled and found negative before the main lung resection. In those patients, going on to perform a full dissection did not improve survival. Median survival was about 8.1 years with sampling versus 8.5 years with dissection, and five-year disease-free survival was essentially identical at 69% and 68%.²³ The catch is that this applied only to early-stage patients whose pre-resection sampling was negative and thorough. The results are not generalizable to patients staged only by imaging or those with more advanced tumors.

A meta-analysis combining the available evidence confirmed the overall picture: for early-stage non-small cell lung cancer, dissection and sampling showed no significant difference in overall survival, local recurrence, distant metastasis, or total complications.²⁴ The practical takeaway is that full dissection remains important for accurate staging and for patients where there is any doubt about node status, but in truly early-stage patients with negative sampling, it may not add survival benefit.

Immunotherapy-Related Lymph Node Swelling

A newer diagnostic trap involves immune checkpoint inhibitors, which are now widely used to treat melanoma, liver cancer, lung cancer, and other malignancies. These drugs can trigger sarcoidosis-like reactions that cause mediastinal lymph nodes to enlarge and light up intensely on PET scans, closely mimicking cancer progression. In one case, a patient with hepatocellular carcinoma on immunotherapy developed multiple enlarged mediastinal nodes that looked like metastatic disease on CT. Biopsy revealed non-necrotizing granulomas, and tumor markers were actually declining, confirming the enlargement was an immune side effect rather than cancer spread.²⁵

Similar cases have been documented with ipilimumab in melanoma patients, where PET-positive mediastinal and hilar nodes turned out to be drug-induced granulomatous inflammation that regressed on its own over several months.²⁶ Recognizing this possibility is increasingly important because misinterpreting a sarcoidosis-like reaction as disease progression could lead to unnecessary treatment changes, including abandoning an immunotherapy regimen that is actually working.

Artificial Intelligence in Mediastinal Node Assessment

Reading PET/CT scans for mediastinal node involvement is time-consuming and reader-dependent, which has made it a natural target for AI-assisted diagnosis. A deep learning model trained to detect pathological mediastinal nodes on PET/CT achieved a sensitivity of 0.87, comparable to an expert reader, with fewer than one false positive per patient on the scanner it was trained on. Performance dropped substantially when the model was tested on images from a different scanner, but transfer learning (retraining with a small set of images from the new machine) restored sensitivity to 0.88.²⁷

A comparison of several machine learning approaches found that a convolutional neural network achieved about 86% accuracy with an area under the curve of 0.91 for classifying mediastinal nodes as metastatic or benign. That performance was comparable to both classical machine learning methods and human physicians.²⁸ These tools are not replacing radiologists yet, but they signal a direction where automated screening could flag suspicious nodes for closer review, potentially reducing the number of cases that slip through on a busy reading list.