The Core Curriculum: Cardiopulmonary Imaging, 1st Edition (2004)

Chapter 9. Radiology of Mediastinal Masses

Evaluation of the mediastinum is an important part of the interpretation of a chest x-ray (CXR). Saying that it is important is not the same as saying that it is well done. The mediastinum is the giant blind spot of the CXR. Mediastinal masses in particular represent a significant challenge to the diagnostic capabilities of the radiologist. Plain-film analysis of mediastinal masses is considered, with pointers for differential diagnosis. The roles of computed tomography (CT) and magnetic resonance imaging (MRI) are also addressed.

Localizing Mediastinal Masses

Localization of mediastinal masses on CXR is a two-part job. The first part is to determine that a mass is actually mediastinal, and the second part is to place it in the anterior, middle, or posterior mediastinum. Several signs place a mass in the mediastinum. Configuration of the interface of the mass with adjacent lung is sometimes helpful. Parenchymal lung masses are generally almost completely surrounded by lung, forming an acute angle with the mediastinum. Mediastinal masses instead have the shape of extraparenchymal masses, pushing toward lung with resultant obtuse angles (Fig. 9.1). As with lateral extraparenchymal masses, it may be very difficult to distinguish a mediastinal mass from one in medial pleura. Bone destruction (involving spine, ribs, or sternum) resolves the issue, indicating that a medial mass is mediastinal. Bilaterality of abnormality also strongly suggests a mediastinal origin (Fig. 9.2). Conversely, air bronchograms in a lesion indicate that it arose in lung, not mediastinum.

Bilaterality of abnormality in proximity to the thoracic midline suggests a mediastinal origin.

Cardiopericardial abnormalities constitute an important pitfall in diagnosis of mediastinal masses. On the one hand, a soft anterior mediastinal mass such as a thymolipoma may droop down along the heart border (Fig. 9.3), resembling pericardial effusion or cyst (1). On the other hand, a pericardial cyst or neoplasm may extend cephalad into the pericardial recesses, simulating a mediastinal mass. CT may be very helpful in this regard.

The ability of radiologists to localize mediastinal masses via CXR has atrophied because of CT. Certain signs (silhouette sign, hilum overlay and convergence signs, and cervicothoracic and thoracoabdominal signs) (Chapter 3) can be enormously helpful in localizing mediastinal masses. Effect of a mass on adjacent mediastinal structures (trachea, paraspinal line, anterior and posterior junction lines, and ribs) should also be assessed. These observations, in combination, will often be more useful for localizing mediastinal masses than will lateral chest radiographs (Fig. 9.4). Although the advantage of having two right-angle radiographs for evaluating three-dimensional anatomy has previously been emphasized, lateral radiography is frequently not very helpful for mediastinal mass localization.

Differential Diagnosis

Generating a differential diagnosis for a mediastinal mass starts with a classification scheme. The system used by Felson (2) (Fig. 9.5) divides the mediastinum into anterior, middle, and posterior compartments by drawing one line along the front of the trachea and the back of the heart and a second line 1 cm posterior to the anterior margin of the thoracic vertebrae. This system does not classify the superior mediastinum as a separate compartment.

For anterior mediastinal masses, the classic differential diagnosis is the “4 Ts”: thymoma, thyroid, teratoma, and terrible lymphoma (Table 9.1). Further clues can be obtained from the radiographic appearance, the patient’s age, and associated clinical manifestations. For example, mediastinal thyroid is rare if there is not demonstrable direct extension from the neck. Calcification, teeth, and/or fat in a lesion favor teratoma. Lymphoma is the likeliest diagnosis when there is a multilobular mass (Fig. 9.6). Thymoma often (although not always) overlies the aortopulmonary window (Fig. 9.7).

The “4 Ts” delineate the important entities in anterior mediastinal mass differential diagnosis—thymoma, thyroid, teratoma, and terrible lymphoma.

Turning to clinical clues, teratoma is typically a disease of teenagers, whereas thymoma usually affects those aged 40 to 60 years. Hodgkin disease has a bimodal age distribution, particularly affecting patients in their teens or 20s and those over age 50. There are a number of clinical conditions associated with thymoma, including myasthenia gravis (MG), pure red cell aplasia, and hypogammaglobulinemia. MG is the most common of these, occurring in roughly 50% of patients with thymomas; among patients with MG, 10% to 15% have an underlying thymoma.

The classic mnemonic is not all inclusive. There are thymic lesions other than thymoma to consider, for example. Thymic carcinoid demonstrates some of the same features as bronchial carcinoid (Fig. 9.8). Metastatic disease and vascular lesions (such as aneurysms) can occur in any mediastinal compartment. Trauma or nontraumatic mediastinal hemorrhage probably affects the anterior mediastinum more than any other mediastinal compartment (Fig. 9.9). Finally, when parathyroid adenomas are not found in the neck, they usually occur in the anterior mediastinum (Fig. 9.10). Only a small percentage (less than 2%) of normal parathyroid glands are mediastinal, but in patients with previously unsuccessful neck exploratory surgery for primary hyperparathyroidism, the incidence of a mediastinal adenoma rises to 47% (3). Whereas most ectopic parathyroid glands are in the superior aspect of the mediastinum, readily accessible to a neck incision, the percentage of mediastinal glands requiring sternotomy rises to 17% in those being reexplored for hyperparathyroidism.

In patients with prior neck exploration and continued primary hyperparathyroidism, the incidence of mediastinal adenoma is 47%, and 17% of such glands cannot be reached from a neck incision.

Turning to the middle mediastinum (Table 9.2), the key structure that traverses this compartment from top to bottom is the esophagus (Fig. 9.11). As a result, even in the twenty-first century CT is not the only alternative to the CXR for diagnosis and further characterization of middle mediastinal masses. Barium swallow actually is better than CT at delineation of esophageal lesions, which account for many middle mediastinal masses. In fact, it is surprising how many patients with documented esophageal carcinomas have normal or nearly normal chest CTs. Other esophageal masses (such as leiomyomas) are similarly better seen with barium swallow than with CXR or CT.

If a middle mediastinal mass is unrelated to the esophagus, the differential diagnosis includes bronchogenic cyst (Fig. 9.12), lymph node abnormalities (sarcoid, lymphoma, metastases), and vascular lesions (Fig. 9.13). Bronchogenic cyst most often occurs between the carina and the esophagus, but the right lower paratracheal region is not an uncommon alternative location (Fig. 9.14). Among lymph node diseases, tuberculous lymphadenitis has become a more important entity because it is common in patients with acquired immunodeficiency syndrome and frequently involves middle mediastinal lymph nodes. Metastatic disease usually affects lymph nodes in the anterior and/or middle mediastinum. Lung cancer is the most common primary neoplasm to involve mediastinal lymph nodes. Most extrathoracic neoplasms do not commonly metastasize to intrathoracic lymph nodes, but there are exceptions. Head and neck tumors, genitourinary neoplasms (especially testicular and renal cell carcinomas), breast carcinoma, gastric carcinoma, and melanoma are extrathoracic primaries with a predilection for hilar and mediastinal lymph nodes.

Extrathoracic neoplasms that metastasize to intrathoracic lymph nodes include primaries of head and neck, breast, stomach, kidney, and testis, as well as melanoma.

Posterior mediastinal masses (Table 9.3) generally represent neurogenic tumors (neurofibroma, schwannoma, ganglioneuroma, and so on) (Fig. 9.15). Posterior mediastinal masses may grow to incredible sizes. (A thoracic surgeon in our institution notes that the largest intrathoracic masses he has encountered have almost always been posterior mediastinal masses.) Interestingly, in patients with neurofibromatosis, posterior mediastinal masses are nearly as often lateral meningoceles (Fig. 9.16) as they are neurogenic tumors. Experience teaches that masses near the medial lung apices are often posterior mediastinal in origin, even when the angle between lesion and mediastinum is not well assessed (Fig. 9.17). This applies to most medial lesions above the clavicles, at least if they are well outlined by lung (the cervicothoracic sign).

For posterior mediastinal masses in general, other diagnostic considerations include disc space or vertebral body infection, vascular lesions, and extramedullary hematopoiesis (Fig. 9.18). The latter condition usually produces bilateral lobulated paravertebral masses in the lower thoracic region and is commonly associated with hereditary spherocytosis, thalassemia, or other severe congenital anemias.

Computed Tomography of Mediastinal Masses

CT has revolutionized the imaging of mediastinal masses. Compared with CXR, CT has major advantages in detection and characterization of mediastinal masses. These advantages are considered in turn, and the ability of CT to arrive at a specific diagnosis is also addressed.

Detection

CT is considerably better than CXR in detecting mediastinal masses. This has been particularly well studied in patients with MG, where underlying thymoma is an important concern. In an early series of six patients with MG (4), CT detected a thymoma missed by CXR in one and accurately localized to the thymus a second calcified lesion whose CXR location was indeterminate. In a second series of 23 consecutive MG patients who underwent thymectomy regardless of neurologic status or imaging results (5), CT detected all four thymomas that were present; CXR was positive in three of four. At CT there were two false-positive examinations, but there were three false positives at CXR. In a larger series of 57 MG patients (6), 16 of whom had thymoma, CT detected 14 of 16. CXR in these patients was positive in nine, equivocal in two, and normal in five.

CT is indicated to rule out thymoma in MG, even when CXR is normal.

Lesion detection was also addressed in a large series of proven mediastinal masses that had been demonstrated by CT (7). Of 90 patients with concurrent CXR, only 62 (69%) had mediastinal masses that could be detected by the chest radiograph. The rate of detection depended on the mediastinal compartment. Only 48% (10/21) of middle mediastinal masses were detected by CXR, with 67% (20/30) of anterior and 100% (4/4) of posterior mediastinal masses visualized. The smallest mass detected by CXR was a 2 × 2 × 2-cm anterior mediastinal mass caused by sarcoidosis. The largest mass missed by CXR was a 6 × 5 × 7 cm middle mediastinal mass in small cell carcinoma of lung.

CXR detected only 69% of CT-demonstrated mediastinal masses in one series.

Confirmation of Location and Extent

The ability of CT to confirm location and extent of mediastinal abnormality has important consequences in differential diagnosis. It is obvious that when CT precisely localizes a mass that cannot be localized by CXR, this will improve differential diagnosis. It is less obvious (but also true) that the ability of CT to differentiate patients with a single mediastinal mass from those with multiple masses also has important diagnostic implications. Patients with multiple masses are very unlikely to have thymoma or teratoma and are much more likely to have lymphoma, metastatic disease, or sarcoidosis (7). On the other hand, a single middle mediastinal mass virtually excludes lymphoma (7). This does not apply to a single anterior mediastinal mass; lymphoblastic lymphoma presented this way in several patients. CT is much better at distinguishing single from multiple masses than CXR. Of 28 patients with multiple masses at CT and an abnormal CXR, 12 (43%) were thought to have a single mass based on the chest radiograph (7).

Identification of Fat, Water, and Calcium

The ability of CT to identify different tissues by their attenuation characteristics can be very important in diagnosing mediastinal masses.Tables 9.4, 9.5, and 9.6 list the masses that may be fat attenuation, fluid attenuation, or calcified, respectively, on CT. An anterior mediastinal mass containing fat, water, calcium, and/or teeth is a teratoma (8) (Fig. 9.19). Thymoma may have a rim of calcification (Fig. 9.20), whereas thymic cyst is a water attenuation lesion. In the middle mediastinum, bronchogenic cyst is often a water attenuation lesion, although uniform higher attenuation in a bronchogenic cyst is a well-known phenomenon (9) (Fig. 9.21). Low attenuation lymph nodes with enhancing rims are typical of tuberculous lymphadenitis (10). As for posterior mediastinal masses, many neurogenic tumors are relatively low in attenuation (Fig. 9.15C) but not of water attenuation. Calcification may be seen in neurogenic tumors, particularly those containing elements of neuroblastoma. Lateral meningoceles are water attenuation paravertebral masses (Fig. 9.16).

Enhancement Characteristics

Bolus injection of intravenous contrast optimizes CT demonstration of enhancement of intrathoracic structures. This in turn facilitates differentiation of mediastinal from cardiac masses, the demonstration of mediastinal vascular abnormalities, and the recognition of enhancing mediastinal masses (Table 9.7). The demonstration of an enhancing mediastinal mass results in a more specific differential diagnosis (11). Mediastinal thyroid is the most common enhancing mediastinal mass. At the University of Michigan, a referral center for extraadrenal pheochromocytomas because the radionuclide ¹³¹I MIBG was developed there (12), intrathoracic pheochromocytomas accounted for 4 of enhancing mediastinal masses and an aortic body tumor accounted for a fifth (11). These tumors can be lumped together because of their common origin from paraganglionic cells, and it has been suggested that all aortic and carotid body tumors, chemodectomas, glomus tumors, and extraadrenal pheochromocytomas should instead be called paragangliomas.

Mediastinal thyroid and paraganglioma account for most enhancing mediastinal masses, whereas lymphoma and metastatic disease almost never enhance.

Other causes of enhancing mediastinal masses include Castleman disease (Fig. 9.22), parathyroid adenoma (Fig. 9.23), and carcinoid tumor. On rare occasions the pattern of contrast enhancement may suggest a specific diagnosis (13) (Fig. 9.24). Contrast enhancement virtually eliminates lymphoma and metastatic disease from consideration.

Detection of Concurrent Abnormalities

Associated abnormalities in the chest or upper abdomen are common in patients with mediastinal masses, and they definitely affect differential diagnosis. Concurrent axillary lymph node enlargement favors lymphoma, whereas a focal pulmonary mass, focal hepatic or adrenal lesions, or bone lesions suggest metastatic disease (7). A very important clue to the diagnosis of malignant thymoma is focal or multifocal, usually unilateral pleural implants. These can be in either hemithorax, although in our experience they are more common in the left pleural space. Amazingly, in some patients it may be more difficult to detect these implants with CT than with CXR. Various other concurrent abnormalities are demonstrable (e.g., pericardial disease, neck mass, abdominal lymph node enlargement), but they do not narrow the differential diagnosis significantly.

Specific Computed Tomography Diagnoses

With the advantages enumerated above it would be expected that CT would improve one’s ability to reach a specific diagnosis in many patients with mediastinal masses. An elaborate algorithm has been developed that uses the capabilities of CT (14). However, in the previously cited series (7) of proven mediastinal masses demonstrated by CT, specific diagnosis was possible in only 16 of 132 patients. Most patients had metastatic disease or lymphoma. These common soft tissue attenuation masses could not reliably be distinguished from each other (or from sarcoidosis), and it was also generally impossible to distinguish Hodgkin disease from non-Hodgkin lymphoma or to distinguish metastatic lung cancer from other metastatic disease. CT may narrow the differential diagnosis, as previously noted, by demonstrating multiple masses or certain concurrent extramediastinal abnormalities, but specific CT attenuation characteristics were encountered infrequently. A more recent study (15) of anterior mediastinal masses concurred, noting “although CT is better than chest radiography in determining the pathologic diagnosis of an anterior mediastinal mass, CT is still poor at making that prediction with confidence.”

Certain mediastinal masses can be characterized with confidence at CT, but specific diagnosis generally remains an elusive goal.

Current Role of Magnetic Resonance Imaging

Several series have shown that MRI can clearly display mediastinal masses in a manner comparable with CT (16,17). Although MRI has theoretical and real advantages (including multiplanar imaging capability, lack of ionizing radiation, avoidance of intravenous contrast agents or use of less toxic agents, and potential for tissue characterization), to date these are outweighed by considerations such as availability, cost, and, most important, examination time. As a result, CT remains the examination of choice for the mediastinum, with MRI assuming a secondary problem-solving role. This has been the state of affairs since the 1980s, and it is unlikely to change in the near future.

References

1. Rosado-de-Christenson ML, Pugatch RD, Moran CA, et al. Thymolipoma: analysis of 27 cases. Radiology 1994;193:121–126.

2. Felson B. Chest roentgenology. Philadelphia: WB Saunders, 1973.

3. Krudy AG, Doppman JL, Brennan MF, et al. The detection of mediastinal parathyroid glands by computed tomography, selective arteriography, and venous sampling. Radiology 1981;140:739–744.

4. Mink JH, Bein ME, Sukor R, et al. Computed tomography of the anterior mediastinum in patients with myasthenia gravis and suspected thymoma. AJR Am J Roentgenol 1978;130:239–246.

5. Moore AV, Korobkin M, Powers B, et al. Thymoma detection by mediastinal CT: patients with myasthenia gravis. AJR Am J Roentgenol1982;138:217–222.

6. Fon GT, Bein ME, Mancuso AA, et al. Computed tomography of the anterior mediastinum in myasthenia gravis. Radiology1982;142:135–141.

7. Rebner M, Gross BH, Robertson JM, et al. CT evaluation of mediastinal masses. Comput Radiol 1987;11:103–110.

8. Moeller KH, Rosado-de-Christenson ML, Templeton PA. Mediastinal mature teratoma: imaging features. AJR Am J Roentgenol1997;169:985–990.

9. Mendelson DS, Rose JS, Efremidis SC, et al. Bronchogenic cysts with high CT numbers. AJR Am J Roentgenol 1983;140:463–465.

10. Im J, Song KS, Kang HS, et al. Mediastinal tuberculous lymphadenitis: CT manifestations. Radiology 1987;164:115–119.

11. Spizarny DL, Rebner M, Gross BH. Enhancing mediastinal masses: CT evaluation. J Comput Assist Tomogr 1987;11:990–993.

12. Francis IR, Glazer GM, Shapiro B, et al. Complementary roles of CT and ¹³¹I-MIBG scintigraphy in diagnosing pheochromocytoma. AJR Am J Roentgenol 1983;141:719–725.

13. McAdams HP, Rosado-de-Christenson ML, Moran CA. Mediastinal hemangioma: radiographic and CT features in 14 patients. Radiology1994;193:399–402.

14. Feigin DS, Padua EM. Mediastinal masses: a system for diagnosis based on computed tomography. J Comput Tomogr 1986;10:11–21.

15. Ahn JM, Lee KS, Goo JM, et al. Predicting the histology of anterior mediastinal masses: comparison of chest radiography and CT. J Thorac Imag 1996;11:265–271.

16. Aronberg DJ, Glazer HS, Sagel SS. MRI and CT of the mediastinum: comparisons, controversies, and pitfalls. Radiol Clin North Am1985;23:439–448.

17. Von Schulthess GK, McMurdo KC, Tscholakoff D, et al. Mediastinal masses: MR imaging. Radiology 1986;158:289–296.