413 itmig classification of mediastinal compartments and ... › ... › 2017 › 2_mayo › rsna...

Ch

est Im

ag

Ing

413

ITMIG Classification of Mediastinal Compartments and Multidisciplinary Approach to Mediastinal Masses1

Division of the mediastinum into specific compartments is ben-eficial for a number of reasons, including generation of a focused differential diagnosis for mediastinal masses identified on imag-ing examinations, assistance in planning for biopsies and surgical procedures, and facilitation of communication between clinicians in a multidisciplinary setting. Several classification schemes for the mediastinum have been created and used to varying degrees in clinical practice. Most radiology classifications have been based on arbitrary landmarks outlined on the lateral chest radiograph. A new scheme based on cross-sectional imaging, principally multi-detector computed tomography (CT), has been developed by the International Thymic Malignancy Interest Group (ITMIG) and accepted as a new standard. This clinical division scheme defines unique prevascular, visceral, and paravertebral compartments based on boundaries delineated by specific anatomic structures at multidetector CT. This new definition plays an important role in identification and characterization of mediastinal abnormalities, which, although uncommon and encompassing a wide variety of entities, can often be diagnosed with confidence based on location and imaging features alone. In other scenarios, a diagnosis may be suggested when radiologic features are combined with specific clinical information. In this article, the authors present the new multidetector CT–based classification of mediastinal compartments introduced by ITMIG and a structured approach to imaging evalu-ation of mediastinal abnormalities.

©RSNA, 2017 • radiographics.rsna.org

Brett W. Carter, MD Marcelo F. Benveniste, MD Rachna Madan, MD Myrna C. Godoy, MD, PhD Patricia M. de Groot, MD Mylene T. Truong, MD Melissa L. Rosado-de-Christenson, MD Edith M. Marom, MD

Abbreviations: FDG = fluorodeoxyglucose, ITMIG = International Thymic Malignancy In-terest Group, JART = Japanese Association for Research on the Thymus, SUVmax = maximal standardized uptake value

RadioGraphics 2017; 37:413–436

Published online 10.1148/rg.2017160095

Content Codes: 1From the Department of Diagnostic Radiol-ogy, University of Texas MD Anderson Can-cer Center, 1515 Holcombe Blvd, Unit 1478, Houston, TX 77030 (B.W.C., M.F.B., M.G., P.M.d.G., M.T.T.); Department of Radiology, Brigham and Women’s Hospital, Boston, Mass (R.M.); Department of Radiology, Saint Luke’s Hospital of Kansas City, University of Missouri–Kansas City School of Medicine, Kansas City, Mo (M.L.R.d.C.); and Department of Radiol-ogy, Chaim Sheba Medical Center, Tel Aviv, Israel (E.M.M.). Recipient of a Certificate of Merit award for an education exhibit at the 2015 RSNA Annual Meeting. Received April 8, 2016; revision requested July 12 and received August 30; accepted September 20. For this journal-based SA-CME activity, the author M.L.R.d.C. has provided disclosures (see end of article); all other authors, the editor, and the reviewers have disclosed no relevant relationships. Address correspondence to B.W.C. (e-mail: [email protected]).

See discussion on this article by Van Schil and Heyman (pp 436–438).

©RSNA, 2017

After completing this journal-based SA-CME activity, participants will be able to:

■ Identify the boundaries and contents of the mediastinal compartments at mul-tidetector CT as outlined in the ITMIG classification system.

■ Describe the role of imaging modalities such as multidetector CT, MR imaging, and FDG PET/CT in evaluating medias-tinal masses.

■ Understand basic approaches to diag-nosing mediastinal masses on the basis of imaging features and clinical information.

See www.rsna.org/education/search/RG.

SA-CME LEArnInG ObjECTIvES

IntroductionThe mediastinum contains vital vascular and nonvascular structures and organs. Division of the mediastinum into specific compartments has traditionally been valuable in the identification, characterization, and management of various mediastinal abnormalities. Numerous classification systems have been developed and used to varying de-grees by anatomists, surgeons, and radiologists. The most commonly used scheme in clinical practice is the Shields classification system, whereas the traditional Fraser and Paré, Felson, Heitzman, Zylak, and Whitten models are used in radiologic practice (1–7). One scheme has been devised to bridge the gap between the various models (8). Significant differences in the terminology and methods of mediastinal division or compartmentalization among these schemes have resulted in confusion among health care providers and the inability to reliably localize some lesions to a specific mediastinal compartment because of inherent limitations in each classification.

The existing schemes used in radiologic practice represent ar-bitrary nonanatomic divisions of the chest based primarily on the lateral chest radiograph. The lack of a classification scheme based on cross-sectional imaging is problematic, as multidetector computed

This copy is for personal use only. To order printed copies, contact [email protected]

Silvisol

Rectángulo

414 March-April 2017 radiographics.rsna.org

describe mediastinal abnormalities and formulate relevant differential diagnoses.

In 2014, the Japanese Association for Research on the Thymus (JART) developed a four-com-partment multidetector CT–based classification scheme for division of the mediastinal compart-ments, which was derived from a retrospective analysis of 445 nonconsecutive pathologically proved mediastinal masses (11). On the basis of discussions with experts in the field of mediastinal diseases, the International Thymic Malignancy Interest Group (ITMIG) has modified the JART model and introduced a new definition of medias-tinal compartments to be used with cross-sectional imaging and adopted as a new standard (12).

In this article, we describe the new ITMIG mediastinal compartment classification system based on cross-sectional imaging that can be used to accurately localize and characterize mediastinal lesions and assist in the formulation of focused differential diagnoses and management strategies. Specific approaches to evaluation of abnormalities in the prevascular, visceral, and paravertebral com-partments are presented here and primarily based on multidetector CT. It is important to note that these approaches are not intended to include every possible entity that may be encountered in the me-diastinum, but instead to provide a practical and realistic algorithm for the radiologist. Although management strategies such as biopsy and surgery may be included when necessary, detailed treat-ment strategies for individual lesions are beyond the scope of this article.

rationale and MethodologyITMIG has an established method of develop-ing international standards for mediastinal dis-eases, which was used in this instance to develop a practical division of the mediastinum based on cross-sectional imaging. A thorough analysis of the existing literature regarding the various compart-ment schemes was performed, with specific atten-tion given to the model developed by JART. In an effort to develop a standard representing a con-sensus among clinicians and researchers interested in mediastinal diseases, a working group within ITMIG identified and surveyed a multidisci-plinary group of experts in thoracic surgery, medi-cal oncology, diagnostic radiology, and pathology regarding their preferences for specific details regarding a multidetector CT–based compartment model. Information requested included the follow-ing: (a) preference for a three-compartment or a four-compartment model and (b) specific reasons for this preference. After drafting of a proposed multidetector CT–based compartment model by an ITMIG work group, it was further refined by an extended work group and ultimately dissemi-

tomography (CT) and magnetic resonance (MR) imaging are predominantly used for diagnosis, evaluation, and management of mediastinal ab-normalities and a growing number of mediastinal lesions are detected with multidetector CT stud-ies performed for lung cancer screening, cardiac screening, and other purposes (9,10). Therefore, a standardized classification scheme based on multidetector CT is necessary to appropriately

TEAChInG POInTS ■ The lack of a classification scheme based on cross-sectional im-

aging is problematic, as multidetector computed tomography (CT) and magnetic resonance (MR) imaging are predominantly used for diagnosis, evaluation, and management of mediasti-nal abnormalities and a growing number of mediastinal le-sions are detected with multidetector CT studies performed for lung cancer screening, cardiac screening, and other purposes. Therefore, a standardized classification scheme based on mul-tidetector CT is necessary to appropriately describe mediastinal abnormalities and formulate relevant differential diagnoses.

■ In some cases, the multidetector CT appearance of thymic hy-perplasia is not straightforward and a more nodular or bulky configuration may be present, which can mimic a thymic epithelial tumor, lymphoma, or other soft-tissue neoplasm. In this setting, two options are available: (a) follow-up multide-tector CT can be performed in ~3 months to allow a decrease in thymic size or (b) chemical shift MR imaging with in-phase and out-of-phase gradient-echo sequences. Thymic hyperpla-sia and the normal thymus characteristically demonstrate loss of signal on out-of-phase images due to the suppression of fat interspersed between nonneoplastic hyperplastic thymus, whereas thymic epithelial neoplasms, lymphoma, and other soft-tissue malignancies do not demonstrate suppression on out-of-phase images. Use of either of these options ensures that unnecessary biopsies or surgeries can be avoided.

■ Internal heterogeneity resulting in soft-tissue attenuation at multidetector CT may be due to hemorrhagic or protein-aceous components or infection of the bronchogenic cyst; in this setting, MR imaging can be performed to confirm the cystic nature of the lesion, which demonstrates high signal intensity on T2-weighted images regardless of the other con-tents. The presence of hemorrhagic, proteinaceous, or mu-coid content results in variable patterns of signal intensity on T1-weighted images.

■ When a previously stable neurofibroma suddenly increases in size, develops regions of heterogeneity, and/or invades ad-jacent tissues, malignant transformation to a malignant pe-ripheral nerve sheath tumor should be strongly considered. In patients with neurofibromatosis type 1, there is a 10% life-time risk of developing a malignant peripheral nerve sheath neoplasm, and these lesions account for most cancer-related deaths in these patients. FDG PET/CT has demonstrated utility in distinguishing malignant peripheral nerve sheath tumors from benign neurofibromas, with sensitivity of 95% and specificity of 72%.

■ Tuberculosis associated with involvement of the posterior ele-ments, the presence of a large prevertebral and paravertebral soft-tissue component out of proportion to the degree of os-seous destruction, and disk space narrowing enable differentia-tion of tuberculous from pyogenic infection. Additionally, the presence of calcification without new bone formation or sclero-sis strongly suggests the diagnosis of Pott disease of the spine.

RG • Volume 37 Number 2 Carter et al 415

Among the surveyed multidisciplinary group of experts, 72% preferred a three-compartment model, 23% preferred a four-compartment model, and 5% did not have a specific preference. Reasons for selecting one model over another included the following: (a) optimal distinction of disease entities (67%), (b) similarity to what is currently used (63%), (c) anatomic nature (53%), and (d) ease of use (48%). On the basis of this feedback, a three-compartment model was selected as the foundation for the cross-sectional classification scheme developed by ITMIG.

ITMIG Definition of Mediastinal Compartments

The three-compartment cross-sectional imaging model of the mediastinal compartments developed by ITMIG includes prevascular (anterior), visceral (middle), and paravertebral (posterior) compart-ments (Table). Specific compartment boundaries and the anatomic structures they contain can be readily identified at multidetector CT (Fig 1).

Prevascular CompartmentIn the ITMIG classification system, the following boundaries of the prevascular compartment are defined: (a) superiorly, the thoracic inlet; (b) infer-iorly, the diaphragm; (c) anteriorly, the posterior border/cortex of the sternum; (d) laterally, the parietal mediastinal pleura; and (e) posteriorly, the anterior aspect of the pericardium as it wraps around the heart in a curvilinear fashion (Table). Briefly, the thoracic inlet has been defined as a thin plane of tissue outlined by specific osseous struc-tures, including the superior border of the manu-brium anteriorly and inferiorly, the body of the first thoracic vertebra posteriorly and superiorly, and the first pair of ribs and their costal cartilages.

On the basis of these landmarks, the major contents of the prevascular compartment in-clude the thymus, fat, lymph nodes, and the left brachiocephalic vein. Therefore, the most com-mon abnormalities encountered in the prevascu-lar compartment include thymic lesions (cysts, hyperplasia, and malignancies such as thymoma, thymic carcinoma, and neuroendocrine neo-plasms); germ cell neoplasms (which arise from germ cell rest remnants in the mediastinum); lymphoma; metastatic lymphadenopathy; and intrathoracic goiter.

visceral CompartmentIn the ITMIG classification system, the following boundaries of the visceral compartment are de-fined: (a) superiorly, the thoracic inlet; (b) inferi-orly, the diaphragm; (c) anteriorly, the posterior boundaries of the prevascular compartment; and (d) posteriorly, a vertical line connecting a point

nated to the active ITMIG members (~225) for review. The final document, which we present in this article, was approved and adopted as a new standard by the ITMIG members.

Mediastinal classification models have tra-ditionally divided the mediastinum into three or four compartments depending on whether a superior mediastinal compartment is included in the description. Four-division models include superior, anterior, middle, and posterior compart-ments, whereas three-division models include anterior, middle, and posterior compartments. In describing a cross-sectional imaging approach, both types of models have specific advantages and disadvantages. The principal advantages of a four-compartment cross-sectional imaging model include similarity to anatomic and radiologic four-compartment models, the efficacy of such a system as demonstrated by the JART proposal, and the fact that most cases of thyroid goiter are typically located in the superior mediastinum and can be distinguished from other mediastinal masses on this basis.

Significant disadvantages of a four-compart-ment cross-sectional imaging model include the relative complexity compared with a more streamlined three-compartment model, the per-ception that most clinicians and radiologists do not use the existing four-compartment schemes, and several nonanatomic features that limit ap-plicability. For example, the division between the superior and inferior compartments is entirely artificial and nonanatomic; thus, neoplastic, infectious, and inflammatory disease processes can spread unimpeded within these two areas without fascial plane restrictions. Furthermore, neurogenic neoplasms in the posterior aspect of the chest do not respect this arbitrary separation of compartments. These factors limit the imple-mentation and dissemination of any four-com-partment cross-sectional imaging model.

The primary advantages of a three-compart-ment cross-sectional imaging model include similarity to the published anatomic, clinical, and radiologic three-compartment models previously developed and in current use, less complicated design relative to four-compartment models, and the fact that specific compartmental boundaries are established along true anatomic planes. The principal disadvantage of a three-compartment cross-sectional imaging model is that blending the superior and anterior compartments may not result in sufficient separation of lesions that occur in each of these locations. However, this is not a significant issue in clinical practice, as specific lesions that typically occupy the superior medias-tinum, in particular thyroid goiter, can be readily identified at multidetector CT.


on the thoracic vertebral bodies 1 cm posterior to the anterior margin of the spine—this is referred to as the visceral-paravertebral compartment boundary line (Table). This vertical line was selected as the posterior boundary of the visceral compartment and the anterior boundary of the paravertebral compartment because most abnor-malities in the latter are neurogenic neoplasms that arise from the dorsal root ganglia/neurons adjacent to the intervertebral foramina.

The major contents of the visceral compart-ment fall into two main categories: (a) vascular structures including the heart, superior vena cava, ascending thoracic aorta, aortic arch, descend-ing thoracic aorta, intrapericardial pulmonary arteries, and thoracic duct; and (b) nonvascular structures including the trachea, carina, esopha-gus, and lymph nodes. In contrast to the JART model, the ITMIG model includes all structures within the pericardium (ie, heart and great ves-sels) in the visceral compartment. Note that the extrapericardial pulmonary arteries and veins are considered to be pulmonary structures and not mediastinal in location; thus, these are not included in the visceral compartment.

The most common abnormalities in the visceral compartment include lymphadenopathy (related to lymphoma or metastatic disease), duplication cysts, tracheal lesions, and esophageal neoplasms. Additionally, vascular lesions arising from the heart, pericardium, and great vessels may also be present.

Paravertebral CompartmentIn the ITMIG classification system, the follow-ing boundaries of the paravertebral compart-ment are defined: (a) superiorly, the thoracic inlet; (b) inferiorly, the diaphragm; (c) ante-riorly, the posterior boundaries of the visceral compartment; and (d) posterolaterally, a vertical line along the posterior margin of the chest wall at the lateral aspect of the transverse processes (Table). The major contents of the paraverte-bral compartment include the thoracic spine and paravertebral soft tissues; therefore, most abnormalities in this region are neurogenic neo-plasms that arise from the dorsal root ganglia/neurons adjacent to the intervertebral foramina. Other potential lesions in this compartment are of infectious (discitis/osteomyelitis) or traumatic (hematoma) origin, or miscellaneous lesions related to other underlying conditions (such as extramedullary hematopoiesis).

Approach to Mediastinal Masses

General ConsiderationsEmploying the existing nomenclature regarding the mediastinal compartments, slightly more than half of all mediastinal masses are located in the anterior compartment, whereas one-fourth each are identified in the middle and posterior medias-tinal compartments (10,13–21). These compart-ments and the frequency of lesions located therein approximate the ITMIG compartment scheme. In

ITMIG Multidetector CT–based Classification of Mediastinal Compartments

Compartment Boundaries Major Contents

Prevascular Superior: thoracic inletInferior: diaphragmAnterior: sternumLateral: parietal mediastinal pleuraPosterior: anterior aspect of the pericardium

as it wraps around the heart in a curvilinear fashion

ThymusFatLymph nodesLeft brachiocephalic vein

Visceral Superior: thoracic inletInferior: diaphragmAnterior: posterior boundaries of the prevascu-

lar compartmentPosterior: vertical line connecting a point on

each thoracic vertebral body 1 cm posterior to its anterior margin

Nonvascular: trachea, carina, esophagus, lymph nodes

Vascular: heart, ascending thoracic aorta, aor-tic arch, descending thoracic aorta, superior vena cava, intrapericardial pulmonary arter-ies, thoracic duct

Paravertebral Superior: thoracic inletInferior: diaphragmAnterior: posterior boundaries of the visceral

compartmentPosterolateral: vertical line against the posterior

margin of the chest wall at the lateral margin of the transverse process of the thoracic spine

Thoracic spineParavertebral soft tissues


Figure 1. ITMIG definition of mediastinal compartments. Sagittal reformatted multidetector CT image (a) and axial multidetector CT images at the levels of the aortic arch (b), left pulmonary artery (c), and left atrium (d) demonstrate the classification scheme. Note that the prevascular compartment (purple) wraps around the heart and pericardium, which are located in the visceral compartment (blue). Yellow = paravertebral compartment, green line = visceral-paravertebral compartment boundary line.

many instances, localization and characterization of a mediastinal abnormality using multidetector CT are sufficient to make the diagnosis. In other cases, correlation between imaging findings and clinical context, as well as additional imaging examina-tions such as MR imaging and fluorodeoxyglucose (FDG) positron emission tomography (PET)/CT and histologic sampling through image-guided or surgical biopsy, is necessary to make a definitive diagnosis and guide further management.

role of Imaging

Radiography.—Although the ITMIG classifica-tion of mediastinal compartments is applicable only to multidetector CT and other cross-sectional imaging modalities, it is important to

consider findings suggestive of a mediastinal abnormality at chest radiography, as it remains the most common imaging examination per-formed. Although small lesions may not be visible or produce subtle findings, large medi-astinal abnormalities may manifest in a variety of ways, such as a soft-tissue mass, often ac-companied by loss of the normal mediastinal contours or interfaces or thickening of specific lines or stripes. The lateral chest radiograph can be especially useful in detecting lesions that may not be visible on the posteroanterior radiograph, since mediastinal lesions may be visible only in the retrosternal space or overlying the upper thoracic spine.

The “silhouette sign,” which describes the loss of normal borders of intrathoracic structures,


can aid in detection of mediastinal abnormalities. For instance, a lesion in the right aspect of the anterior mediastinum may obscure cardiovascular structures such as the superior vena cava or right heart border, whereas a mass in the posterior mediastinum may result in loss of the normal paraspinal stripes. The “hilum overlay” sign may help differentiate a mediastinal mass from cardio-megaly or enlarged pulmonary vessels.

The “cervicothoracic sign” as described by Fel-son (4) is useful for localizing mediastinal abnor-malities identified at radiography and for formulat-ing a focused differential diagnosis. In such cases, obscuration of the lateral borders of an upper mediastinal mass as it extends above the clavicles into the neck implies that the lesion has both intra-thoracic and cervical components. Although often misinterpreted as a manifestation of the cervico-thoracic sign, an upper paravertebral (or posterior mediastinal) mass with borders visible above the clavicles is located entirely within the chest but does not exhibit the cervicothoracic sign.

Multidetector CT.—Once an abnormality is identified at chest radiography, cross-sectional imaging is employed to characterize the lesion, formulate a focused differential diagnosis, as-sess for other abnormalities, and guide further management. Multidetector CT with intrave-nous contrast material is the imaging modality of choice for evaluation and characterization of most mediastinal lesions. One study that analyzed 127 anterior mediastinal masses from various causes demonstrated that multidetector CT was equal or superior to MR imaging in diagnosis of anterior mediastinal masses except for thymic cysts (22). For this reason, ITMIG uses multide-tector CT as the gold standard/reference modal-ity for defining the mediastinal compartments.

Specific imaging characteristics that should be noted at multidetector CT include (a) loca-tion, size, and configuration of mediastinal lesions; (b) descriptive features such as attenua-tion, heterogeneity, and enhancement; (c) pres-ence of intralesional fat, cystic components, soft tissue, and calcification; and (d) any connection with or invasion of adjacent structures. Some of these findings are more important than others; for instance, the presence of fat in a prevascular mediastinal lesion is highly suggestive of a small number of entities, whereas calcifications—whether punctate, coarse, or curvilinear—are nonspecific and cannot be used to discriminate benign from malignant prevascular mediastinal masses, as they may be associated with malig-nant neoplasms such as thymoma or treated lymphoma, as well as benign lesions such as mature teratoma (10,23).

MR Imaging.—MR imaging is not typically performed for evaluation of all mediastinal ab-normalities; however, its effectiveness in specific scenarios has been demonstrated. For instance, MR imaging is the most useful imaging modality for distinguishing cystic from solid lesions (eg, thymic cysts from solid neoplasms), discerning cystic and/or necrotic components within solid masses, distinguishing cystic neoplasms from be-nign cysts, and identifying septa and/or soft tissue within cystic lesions (24,25).

For patients unable to undergo contrast-en-hanced multidetector CT due to renal failure or allergy to intravenous contrast material, nonen-hanced MR imaging with specific fluid-sensitive sequences may be performed to characterize the lesion and evaluate for involvement of vascular structures (26). Chemical shift MR imaging using in-phase and out-of-phase sequences is the best modality for differentiating thymic hyperplasia from thymoma and other thymic neoplasms in adult patients (27,28). Thymic hyperplasia can be identi-fied when the microscopic fat interspersed between thymic tissue loses signal on out-of-phase images.

FDG PET/CT.—The role of fluorine 18 FDG PET/CT in the evaluation of many mediastinal ab-normalities remains controversial. Several stud-ies have been performed to investigate the ability of PET/CT to allow distinction between benign and malignant mediastinal lesions and between various types of malignant primary mediastinal neoplasms. In one of these, malignant neoplasms demonstrated significantly higher FDG uptake than benign lesions when a maximal standardized uptake value (SUVmax) equivalent value of 3.5 was used as a threshold (29). Others have used higher SUVmax thresholds such as 4.67 and suggested that PET/CT is complementary to other conventional imaging techniques and could potentially avoid un-necessary investigations, but noted that histologic sampling is required to confirm PET findings (30). There is significant overlap between the SUVmax of malignant neoplasms demonstrating increased FDG uptake, particularly high-risk thymic epithe-lial neoplasms (World Health Organization [WHO] types B2 and B3), lymphoma, paraganglioma, and nonseminomatous germ cell neoplasms (31).

Regarding thymic epithelial neoplasms, Sung et al (32) suggested that PET/CT could be used to distinguish low-risk thymomas (WHO types A, AB, and B1) from thymic carcinoma. Other groups have reported that PET/CT could be used to distinguish low-risk thymomas from high-risk thymomas (WHO types B2 and B3) and thymic carcinoma (33). However, other studies have been less definitive, with PET/CT not demonstrating a significant benefit in the staging of patients with


thymic epithelial neoplasms. It has been observed that thymic epithelial neoplasms tend to demon-strate variable, often only low-grade FDG uptake, making histologic differentiation between the vari-ous types of neoplasms unreliable.

A significant factor limiting the ability of PET/CT to accurately characterize mediastinal lesions is the potential for false-positive and false-negative examinations. Some benign processes such as thy-mic hyperplasia and inflammatory diseases such as fibrosing mediastinitis may demonstrate increased FDG uptake and mimic malignancy. Jerushalmi et al (34) demonstrated that FDG uptake in thymic hyperplasia is highly variable and in many instances can significantly overlap with that of mediastinal malignancies, with SUVmax as high as 7.3. In such cases, a combination of clinical history, focality of FDG uptake at PET/CT, and morphologic features at multidetector CT is necessary to deter-mine whether the lesion is benign or malignant.

Localization of Mediastinal AbnormalitiesAlthough localization of mediastinal lesions to a specific compartment is an important component of characterization, this may be difficult in some instances. For example, a large mediastinal lesion may appear to involve multiple compartments or extend from one compartment to another, making identification of the precise site of origin challeng-ing. Two tools have been described by ITMIG and are recommended to help identify the compart-ment from which these lesions originate.

One of these tools is known as the “center method” and states that the center of a medias-tinal lesion, defined as the center point of the le-sion on the axial CT image that demonstrates the largest size of the abnormality, localizes the lesion to a specific mediastinal compartment (12). The JART study used this method and resulted in ac-curate localization of all 445 mediastinal masses in the study to specific compartments.

The second tool is known as the “structure dis-placement tool” and is useful in scenarios in which very large mediastinal lesions displace organs from other mediastinal compartments, typically those that abut the compartment from which the lesion originated (11). For instance, a very large prevas-cular mediastinal mass may posteriorly displace organs of the visceral mediastinal compartment such as the trachea, esophagus, or heart.

Approach to the Prevascular Compartment

General ConsiderationsThe true incidence of prevascular mediastinal masses is difficult to ascertain for multiple rea-

sons, the most significant of which is that different classification schemes have been used to define the mediastinal compartments in published stud-ies (35). Additionally, there is variability in the inclusion of nonneoplastic lesions such as thymic hyperplasia and thymic and pericardial cysts and other neoplastic lesions such as lymphoma (35).

The most common neoplasms of the pre-vascular mediastinum include thymic epithelial neoplasms (thymoma, thymic carcinoma, and thymic neuroendocrine tumors) and lymphoma. Thymoma is the most common prevascular mediastinal mass and primary neoplasm of the prevascular mediastinum, with the highest inci-dence in middle-aged patients. Other neoplasms that may arise in the prevascular compartment include mature teratoma, nonteratomatous germ cell malignancies such as seminoma and nonsem-inomatous germ cell neoplasms, and metastatic disease. Nonneoplastic lesions of the prevascular mediastinum include substernal extension of thyroid goiter, thymic hyperplasia, cystic lesions such as thymic and pericardial cysts, and vascu-lar-lymphatic abnormalities.

Lesions Identifiable at Imaging

Thyroid Goiter.—A heterogeneous prevascular mediastinal mass that demonstrates continuity with the cervical thyroid gland, is intrinsically hyperattenuating (with Hounsfield unit values of 70–85 due to the presence of iodine), and demonstrates intense and sustained enhance-ment after administration of intravenous con-trast material can dependably be diagnosed as a mediastinal goiter. Cystic changes manifesting as internal foci of low attenuation and calcifi-cations may be present. For cases in which a definitive connection with the cervical thyroid gland cannot be identified, these other findings at multidetector CT should strongly suggest the diagnosis (Fig 2). When additional findings such as loss of mediastinal tissue planes or associ-ated cervical or mediastinal lymphadenopathy accompany a mediastinal goiter, then thyroid malignancy should be suspected and further evaluation performed (10,36).

Fat-containing Lesions.—The presence of visible regions of intralesional fat measuring between −40 and −120 HU at multidetector CT within a heterogeneous prevascular mediastinal mass is highly suggestive of a mature teratoma. These be-nign lesions characteristically demonstrate vary-ing amounts of fat, fluid, calcification, and soft tissue; in rare cases, bone and tooth-like elements may be identified (23,37) (Fig 3a). Intralesional fat is identified in 50% of cases (37). Fat-fluid


Figure 3. Fat-containing masses. (a) Mature teratoma in a 42-year-old man. Coned-down axial contrast-en-hanced multidetector CT image at the level of the pulmonary arteries shows a large, lobular, heterogeneous prevascular mass with a large macroscopic fat component (M). Regions of soft tissue and fluid (*) are also pres-ent. These findings are virtually pathognomonic for a mature teratoma, which was confirmed at surgical resec-tion. (b) Thymolipoma in a 43-year-old man. Coned-down axial contrast-enhanced multidetector CT image at the level of the left atrium shows a predominantly fat-containing mass (M) in the left prevascular mediastinum. Minimal internal soft tissue is present (arrows), and the lesion extended inferiorly into the left cardiophrenic angle (not shown). When a large fat-containing lesion is identified in a cardiophrenic angle with scattered re-gions of solid tissue and/or fibrous septa, thymolipoma should be considered. Establishment of a connection to the anatomic location of the thymus supports the diagnosis.

levels are highly specific for mature teratoma but are much less common (38). These neoplasms typically affect young patients, accounting for 25% of prevascular masses in patients 10–19 years of age, 10%–15% in patients 20–49 years of age, and less than 5% in patients over 50 years of age in both men and women (10).

Other fat-containing lesions originating from the prevascular mediastinum are much less com-mon and include a wide variety of benign and ma-lignant neoplasms such as thymolipoma, lipoma, and liposarcoma. A predominantly fat-containing lesion present in a cardiophrenic angle may

represent a thymolipoma, an uncommon lesion accounting for less than 5% of prevascular medi-astinal masses in all age groups (Fig 3b). These benign encapsulated neoplasms are comprised of 50%–85% fat—although a fat composition of up to 95% has been reported—and scattered regions of solid tissue and fibrous septa (37,39). Thymo-lipomas tend to be very large at presentation with an average reported size of 20 cm, and a direct connection with the anatomic location of the thymus may be identified. Patients may be asymp-tomatic or have symptoms related to local mass effect; rare associations between thymolipoma and

Figure 2. Thyroid goiter in an 83-year-old woman. Coned-down axial contrast-enhanced multide-tector CT image of the upper chest below the thoracic inlet shows a heterogeneous mass with enhanc-ing soft tissue (black *) and cystic components (white *) in the pre-vascular mediastinum. Note the focus of calcification anteriorly (arrow). Although a definitive con-nection with the cervical thyroid gland could not be identified, these multidetector CT findings strongly suggest the diagnosis of mediastinal goiter.


Figure 4. Thymic cyst in a 43-year-old man. (a) Coned-down axial contrast-enhanced multidetector CT im-age below the level of the carina shows a well-defined low-attenuation lesion (arrow) in the thymic bed of the prevascular mediastinum. The internal Hounsfield unit value was 15. (b) Coned-down axial T2-weighted MR image shows high signal intensity in the lesion (arrow) and no evidence of internal septa or soft tissue. These findings are compatible with a benign unilocular thymic cyst. MR imaging is the most useful imaging modality for distinguishing cystic from solid lesions, discerning cystic and/or necrotic components within solid masses, and identifying septa and/or soft tissue in cystic lesions.

myasthenia gravis, Graves disease, and hemato-logic disorders have been reported (40).

Lipomas represent 2% of all primary mediasti-nal neoplasms and manifest at multidetector CT as encapsulated lesions composed predominantly of fat with a small amount of soft tissue and blood vessels in the prevascular compartment. Although mediastinal liposarcomas are also comprised predominantly of fat, these lesions may be distin-guished from lipomas, thymolipomas, and other fat-containing lesions by the presence of aggressive features, including a greater proportion of soft-tissue components, local invasion, intrathoracic lymphadenopathy, and metastatic disease (41,42).

Fat-containing masses with attenuation values overlapping with those of other tissues at multi-detector CT can be further evaluated with MR imaging. Fat demonstrates high signal intensity on T1- and T2-weighted images, low signal inten-sity on fat-saturated images, and loss of signal on out-of-phase images (37).

Cystic Lesions.—Cystic lesions of the mediasti-num are those that have water or fluid attenua-tion at multidetector CT, with Hounsfield unit values between 0 and 20. A well-circumscribed homogeneous lesion in the prevascular medias-tinum near the thymic bed that is round, oval, or saccular likely represents a thymic cyst. Most of these lesions are acquired and due to inflam-mation; iatrogenic processes such as surgery, radiation therapy, or chemotherapy; or malignant neoplasms. Some thymic cysts may demonstrate

regions of higher attenuation than that of fluid due to hemorrhagic or proteinaceous compo-nents. In such a scenario, MR imaging should be performed, given its ability to allow distinction between cystic and solid lesions and identifica-tion of solid internal components.

Prevascular masses that are purely cystic with no soft-tissue components or internal septa can reliably be diagnosed as unilocular thymic cysts (43) (Fig 4). However, when cystic lesions contain internal soft-tissue components and/or internal septa, the differential diagnosis should include multilocular thymic cysts, cystic tera-toma, lymphangioma, and cystic thymoma. In the clinical setting of symptoms related to myasthenia gravis or other paraneoplastic syndromes, espe-cially in men and women older than 40 years, the diagnosis of cystic thymoma should be strongly considered (Fig 5).

In the setting of a large multilocular cystic lesion with internal septa and/or soft-tissue components that extends into the neck, axilla, or chest wall, the diagnosis of lymphangioma should be considered. Although mature teratomas may demonstrate internal fat at multidetector CT, a large percentage of these lesions manifest as predominantly or entirely unilocular or multiloc-ular thin-walled cystic masses in the prevascular mediastinum. In contrast to simple cysts, cystic teratomas are typically associated with additional features such as internal septa and soft-tissue components and may enhance after administra-tion of intravenous contrast material.


Two distinct histologic types of thymic hy-perplasia have been described: true and thymic lymphoid (follicular) hyperplasia. In patients who have been treated with chemotherapy, radiation therapy, or corticosteroids or exposed to stresses such as burns or injuries, true thymic hyperplasia should be considered when there is diffuse sym-metric enlargement of the thymus at multidetector CT (10). True thymic hyperplasia is also known as “rebound hyperplasia” and is characterized by an increase in thymic volume of greater than 50% over baseline after a causative stressor; approxi-

Figure 5. Cystic thymoma in a 59-year-old woman. (a) Coned-down axial contrast-enhanced multidetector CT image at the level of the pulmonary valve shows a lobular low-attenuation mass (arrow) in the right pre-vascular mediastinum. Although a thymic cyst was suspected on the basis of the CT appearance, the internal Hounsfield unit value was 45. (b, c) Coned-down axial fat-saturated T2-weighted (b) and coned-down axial contrast-enhanced T1-weighted (c) MR images confirm that portions of the lesion are cystic (arrow in b); how-ever, peripheral enhancement (arrows in c) and an internal enhancing soft-tissue septum (arrowhead in c) are identified. These imaging findings, in combination with the clinical history of myasthenia gravis, were highly suggestive of cystic thymoma, which was confirmed at surgical resection. A cystic lesion containing internal soft-tissue components and/or internal septa may represent a multilocular thymic cyst, lymphangioma, cystic teratoma, or cystic thymoma. In the clinical setting of symptoms related to myasthenia gravis or other paraneo-plastic syndromes, as in this case, the diagnosis of cystic thymoma can reliably be made.

A well-circumscribed unilocular mass of fluid attenuation with thin or imperceptible walls local-ized to a cardiophrenic angle can be confidently diagnosed as a pericardial cyst (Fig 6). These benign nonneoplastic lesions arise from aber-rations in the formation of coelomic or somatic cavities and more commonly arise in the right costophrenic angle than in the left, although they may be seen as high as the pericardial recesses at the level of the proximal aorta and pulmonary ar-teries (44). Although pericardial cysts are always connected to the pericardium, only a small num-ber of cases demonstrate this communication at the time of surgery.

Lesions Identifiable with a Combination of Imaging and Clinical Context

Thymic Hyperplasia.—Normal prevascular thy-mic tissue is seen in young individuals and de-creases in prominence with advancing age, with complete fatty replacement usually achieved by 40 years of age. Thymic hyperplasia should be considered in young patients with uniform en-largement of the thymus that is new compared with the appearance on earlier imaging studies, or in patients greater than 40 years of age with soft tissue in the thymic bed that has a con-figuration similar to that of the normal bilobed triangular thymus.


mately 10%–25% of patients undergoing chemo-therapy may develop rebound hyperplasia (45).

Thymic lymphoid (follicular) hyperplasia is a histologic diagnosis defined as the presence of an increased number of lymphoid follicles, which may or may not be associated with an increase in the size of the gland and is typically associated with immunologic diseases such as myasthenia gravis, hyperthyroidism, collagen vascular diseases, or hu-man immunodeficiency virus (HIV) infection (10). At multidetector CT, follicular hyperplasia results in a normal appearance of the thymus, thymic enlargement, or a focal soft-tissue thymic mass. A rare manifestation of thymic hyperplasia at multi-detector CT is a heterogeneously hypoattenuating prevascular mediastinal mass that results from de-position of fat between hyperplastic thymic tissue.

In some cases, the multidetector CT appearance of thymic hyperplasia is not straightforward and a more nodular or bulky configuration may be pres-ent, which can mimic a thymic epithelial tumor, lymphoma, or other soft-tissue neoplasm. In this setting, two options are available: (a) follow-up multidetector CT can be performed in ~3 months to allow a decrease in thymic size or (b) chemical shift MR imaging with in-phase and out-of-phase gradient-echo sequences. Thymic hyperplasia and the normal thymus characteristically demonstrate loss of signal on out-of-phase images due to the suppression of fat interspersed between nonneo-plastic hyperplastic thymus, whereas thymic epi-thelial neoplasms, lymphoma, and other soft-tissue malignancies do not demonstrate suppression on out-of-phase images (27,28) (Fig 7). Use of either of these options ensures that unnecessary biopsies or surgeries can be avoided.

Thymic Epithelial Neoplasms.—A solid, homo-geneous or slightly heterogeneous mass in the prevascular mediastinal compartment in men and

women older than 40 years with immunologic diseases such as myasthenia gravis (most com-mon) or other paraneoplastic syndromes such as pure red cell aplasia/Diamond-Blackfan syn-drome, hypogammaglobulinemia, or aplastic ane-mia likely represents a thymoma (46) (Fig 8a). More than 80% of thymomas can be accurately diagnosed with multidetector CT or MR imag-ing in this setting, and tissue diagnosis is typically unnecessary (22). Pleural or pericardial dissemi-nation may be present in advanced disease and is highly suggestive of thymoma; however, lymph-adenopathy is typically absent.

A thymic epithelial neoplasm other than thymoma, such as thymic carcinoma or thymic carcinoid, should be considered when a large soft-tissue mass is present in the prevascular mediastinum with accompanying features such as increased heterogeneity, local invasion, lymph-adenopathy, and/or distant metastasis (Fig 8b). Specifically regarding thymoma, studies have shown that lobulated or irregular contours, cystic or necrotic regions within the lesion, and multifo-cal calcifications are more suggestive of invasive thymoma than noninvasive thymoma (47,48).

Lymphoma.—A mildly enhancing lobular soft-tissue mass or group of enlarged lymph nodes in the prevascular mediastinum at multidetector CT, especially in the setting of lymphadenopathy in the neck, axilla, or elsewhere in the body, may represent lymphoma. Primary mediastinal lymphoma may be due to Hodgkin lymphoma or non-Hodgkin lym-phoma, encompassing entities such as diffuse large B-cell lymphoma, gray zone lymphoma, and T-cell lymphoblastic lymphoma. Enlarged lymph nodes, often affecting several different lymph node groups or stations, without a focal mediastinal mass suggest secondary involvement by non-Hodgkin lymphoma arising from another location.

Figure 6. Pericardial cyst in a 48-year-old man. Coned-down axial contrast-enhanced multidetector CT image at the level of the ventricles shows a well-circumscribed, ho-mogeneous, fluid attenuation lesion (M) with imperceptible walls in the right cardiophrenic angle. These findings are pathognomonic for a pericardial cyst.


Figure 7. Thymic hyperplasia in a 34-year-old woman with hyperthyroidism. (a) Coned-down axial contrast-enhanced multidetector CT image below the level of the aortic arch shows a lobular soft-tissue mass (M) in the thymic bed of the prevascular mediastinum. (b, c) Axial in-phase (b) and out-of-phase (c) T1-weighted MR images show complete loss of signal in the mass (M) on the out-of-phase image, indicating the presence of mi-croscopic fat interspersed between hyperplastic thymic tissue in thymic hyperplasia. In some cases, the multide-tector CT appearance of thymic hyperplasia is not straightforward and a nodular or bulky configuration may be present, as in this case. Chemical shift MR imaging with in-phase and out-of-phase gradient-echo sequences can be used to differentiate thymic hyperplasia and normal thymus from thymic epithelial neoplasms, lymphoma, and other soft-tissue malignancies, as the latter do not show suppression on out-of-phase images. Use of this imaging technique ensures that unnecessary biopsies or surgeries can be avoided.

Although it may be difficult to distinguish lymphoma from other soft-tissue mediastinal masses, the infiltrative nature of some types of lymphoma enables differentiation from thymic epithelial neoplasms and germ cell tumors. In many cases, lymphomas encase or encircle vascular structures but do not result in invasion. When such findings are present in young pa-tients who present with “B” symptoms such as fever, weight loss, and night sweats, which occur in ~50% of cases, the diagnosis of mediastinal lymphoma can be reliably made (Fig 9). Fur-ther evaluation is typically performed with core

needle biopsy combined with aspiration for flow cytometry or surgical biopsy.

FDG PET/CT has become the modality of choice for staging and restaging of disease for many types of lymphoma, as it is more accurate than multidetector CT in detecting involvement of lymph nodes, with sensitivity of 94% and specificity of 100% compared with 88% and 86%, respectively, for multidetector CT (Fig 9). PET/CT is also effective in identifying intra-nodal and extranodal disease within the body, with sensitivity of 88% and specificity of 100% compared with 50% and 90%, respectively,


Figure 8. Thymic epithelial neoplasms. (a) Thymoma in a 64-year-old man. Coned-down axial contrast-en-hanced multidetector CT image at the level of the left pulmonary artery shows a homogeneous soft-tissue mass (M) in the prevascular mediastinum. The clinical presentation was significant for progressive worsening of myasthenia gravis. This combination of clinical and imaging information enabled prospective diagnosis of a thymoma. Thymoma should be strongly suspected when a prevascular soft-tissue mass is identified in a patient older than 40 years and in the clinical setting of myasthenia gravis or another paraneoplastic syndrome such as pure red cell aplasia/Diamond-Blackfan syndrome or hypogammaglobulinemia. (b) Thymic carcinoma in a 52-year-old man. Coned-down axial contrast-enhanced multidetector CT image at the level of the aortic arch shows a heterogeneous soft-tissue mass (M) in the prevascular mediastinum, ipsilateral and contralateral mediastinal lymphadenopathy (*), and pleural metastatic disease in the right hemithorax (arrow). CT-guided biopsy revealed advanced thymic carcinoma. When a prevascular mediastinal mass is accompanied by features such as local invasion, lymphadenopathy, or distal metastasis, a diagnosis other than thymoma, such as thymic carcinoma or thymic carcinoid, should be considered.

for multidetector CT for detecting end-organ involvement (49).

Nonteratomatous Germ Cell Neoplasms.—Non-teratomatous germ cell neoplasms include a wide variety of lesions, the most common of which include seminoma and nonseminomatous germ cell tumors (NSGCTs). These lesions typically manifest as large soft-tissue masses in the pre-vascular mediastinum, and it can be difficult to distinguish these entities from lymphoma. How-ever, the addition of demographic and clinical/serologic information usually enables diagnosis.

The presence of a large, lobular, homoge-neous soft-tissue mass in the prevascular me-diastinum at multidetector CT in men 10–39 years of age should raise suspicion for seminoma (50) (Fig 10a). Approximately 10% of patients with seminoma demonstrate slightly elevated serum levels of b–human chorionic gonadotro-pin (b-HCG); however, a-fetoprotein (a-FP) level is usually normal. Although serum lactate dehydrogenase (LDH) levels are usually el-evated, this may be seen with other malignancies such as lymphoma (51,52). Pleural effusions are rare but pulmonary metastases are relatively common, which also helps distinguish semi-noma from many types of lymphoma. Further

evaluation with core needle or surgical biopsy is usually performed.

When a heterogeneous prevascular mediastinal mass is present with lung metastases in men below the age of 40 years, NSGCTs should be included in the differential diagnosis (23,53) (Fig 10b). Ap-proximately 90% of patients present with markedly elevated serum levels of a-FP or b-HCG, which is often highly suggestive of the disease (54,55).

Ectopic Parathyroid Adenoma.—In a patient with a clinical history of primary hyperparathy-roidism, elevated serum calcium levels and/or ele-vated serum parathyroid hormone, with or without prior surgical parathyroidectomy, and a soft-tissue nodule in the prevascular mediastinum at multide-tector CT, an ectopic parathyroid adenoma should be suspected. Although most parathyroid adeno-mas are juxtathyroid in location, they may occur in an ectopic location, with the mediastinum being the most common such site (56).

Many different imaging modalities can assist in diagnosis of these lesions, including high-reso-lution ultrasonography (US) with color Doppler, technetium 99m (99mTc) sestamibi single photon emission CT (SPECT), multidetector CT, and MR imaging. Recently, four-dimensional (4D) multidetector CT has been shown to be more


sensitive than US and scintigraphy for preopera-tive identification and localization of parathyroid adenomas. The typical findings at 4D multidetec-tor CT include intense contrast enhancement in the arterial phase, washout of contrast material in the delayed phase, and low attenuation at nonen-hanced multidetector CT (57,58). An enlarged feeding artery or draining vein, referred to as the polar vessel, associated with the hypervascular parathyroid adenoma may also be seen (59).

Approach to the visceral Compartment

General ConsiderationsAs the visceral compartment contains both nonvascular and vascular structures, a wide variety of abnormalities may originate from this region. The most significant lesions in-clude neoplasms of the airways, esophagus, and lymph nodes; enhancing masses; and non-neoplastic abnormalities such as bronchogenic

and esophageal duplication cysts. A discussion of all visceral compartment abnormalities is beyond the scope of this article; the approach described here is based primarily on whether lesions can be identified at imaging alone or with a combination of radiologic and clinical information, and to a lesser extent by organ of origin and composition.

Lesions Identifiable at Imaging

Cystic Lesions.—A well-circumscribed, homoge-neous, fluid attenuation lesion measuring 0–20 HU at multidetector CT in the visceral mediastinal compartment is compatible with a benign dupli-cation cyst, the most common of which include bronchogenic and esophageal duplication cysts. Bronchogenic cysts result from abnormal budding of the primitive foregut, which also gives rise to the tracheobronchial tree, during embryologic develop-ment. Although bronchogenic cysts may arise from any mediastinal compartment, they typically occur

Figure 9. Mediastinal lymphoma in a 20-year-old man who presented with classic “B” symptoms, including fever, night sweats, and weight loss. (a) Coned-down axial contrast-enhanced multide-tector CT image at the level of the pulmonary ar-teries shows a large soft-tissue mass (M) in the pre-vascular mediastinum, resulting in leftward shift of the heart. CT-guided core needle biopsy revealed Hodgkin lymphoma. (b, c) Coned-down axial FDG PET/CT images before (b) and after (c) one cycle of chemotherapy show that the lymphoma is hetero-geneously FDG avid due to the presence of internal necrosis, but decreases markedly in size and FDG uptake on the restaging PET/CT image. FDG PET/CT is the imaging modality of choice for staging and restaging many types of lymphoma, as it is more ac-curate than multidetector CT for detecting lymph node and end-organ involvement.


Figure 10. Nonteratomatous germ cell neoplasms. (a) Seminoma in a 37-year-old man. Coned-down axial contrast-enhanced multidetector CT image at the level of the aortic arch shows a large soft-tissue mass (M) in the prevascular mediastinum. The clinical presentation was significant for elevated serum levels of b–human chorionic gonadotropin (b-HCG) and lactate dehydrogenase (LDH) but normal a-fetoprotein (a-FP) level. This combination of radiologic and clinical information strongly suggested a seminoma, which was confirmed with CT-guided biopsy. (b) NSGCT in a 29-year-old man. Coned-down axial contrast-enhanced multidetector CT image at the level of the pulmonary arteries shows a heterogeneous soft-tissue mass (M) in the right prevascular mediastinum that exerts mass effect on the right side of the heart. Note the presence of right pleural metastasis and a right pleural effusion (arrow). The clinical presentation was significant for elevated serum levels of a-FP, which, in a patient of this age, was suggestive of an NSGCT, as confirmed with CT-guided biopsy.

in the visceral compartment near the carina or, less commonly, the right paratracheal region (44).

At multidetector CT, bronchogenic cysts manifest as a single, smooth, round or ovoid mass with internal low attenuation (Fig 11). The wall has variable perceptibility and may enhance or demonstrate intrinsic calcifications. Internal heterogeneity resulting in soft-tissue attenuation at multidetector CT may be due to hemorrhagic or proteinaceous components or infection of the bronchogenic cyst; in this setting, MR imaging can be performed to confirm the cystic nature of the lesion, which demonstrates high signal intensity on T2-weighted images regardless of the other con-tents. The presence of hemorrhagic, proteinaceous, or mucoid content results in variable patterns of signal intensity on T1-weighted images (Fig 11).

Esophageal duplication cysts are uncommon developmental anomalies that manifest as well-circumscribed, homogeneous, fluid attenuation le-sions adjacent to the esophagus or associated with the esophageal wall at multidetector CT (44) (Fig 12). As with other cystic lesions, internal hetero-geneity may be present, in this case typically due to hemorrhage or infection caused by the presence of ectopic gastric mucosa. The presence of ectopic gastric mucosa can render these lesions visible on 99mTc sodium pertechnetate scans and may be helpful in diagnosing these lesions in pediatric pa-tients, in 50% of whom thoracic duplication cysts contain ectopic gastric mucosa (60). In contrast to

bronchogenic cysts, esophageal duplication cysts may have thick walls.

Lesions Identifiable with a Combination of Imaging and Clinical Context

Enhancing Masses.—Several entities should be considered when a visceral compartment mass demonstrates enhancement after administration of intravenous contrast material. Paragangliomas or extra-adrenal pheochromocytomas are highly vascular neoplasms that arise from chromaffin tissue located in the para-aortic ganglia that may secrete catecholamines; however, the majority of these neoplasms are nonfunctional. Typical clinical symptoms include hoarseness, dysphagia, shortness of breath, and chest pain (61,62). Paragangliomas usually manifest as intensely and homogeneously enhancing mediastinal masses at multidetector CT, although regions of internal heterogeneity repre-senting necrosis may be present (Fig 13a). Further diagnostic studies such as MR imaging, at which paragangliomas demonstrate intermediate signal intensity on T1-weighted images and high signal intensity on T2-weighted images, and iodine 123 (123I) metaiodobenzylguanidine (MIBG) scintigra-phy can help support the diagnosis.

Castleman disease represents a form of non-clonal lymph node hyperplasia and is classified as hyaline vascular, plasma cell, or human herpesvi-rus 8 (HHV-8) types (63) and may be unicentric


or multicentric. Unicentric hyaline vascular type is the most common and may manifest as a non-invasive mass, an infiltrative mass with associated lymphadenopathy, or matted lymphadenopathy with intense enhancement (64) (Fig 13b).

Finally, metastatic disease from vascular pri-mary malignancies such as renal cell and thyroid neoplasms, melanoma, choriocarcinoma, and some sarcomas may result in intensely enhanc-ing lymphadenopathy in the visceral compart-ment. The diagnosis of metastatic lymphadenop-athy can be inferred when the clinical history is significant for one of these neoplasms. In the absence of such information, further evalua-tion with histologic sampling may be necessary; however, this should be done carefully, as biopsy of mediastinal vascular lesions may result in significant intrathoracic hemorrhage.

Esophageal Lesions.—When an abnormality of the thoracic esophagus such as wall thickening or a focal mass is identified at multidetector CT,

the possibility of esophageal malignancy must be considered. Any portion of the esophagus may be affected, although the distal esophagus is typically involved due to the rising incidence of adenocar-cinoma due to gastroesophageal reflux disease. The differential diagnosis for esophageal wall thickening is broad and includes inflammatory, infectious, and neoplastic causes. The presence of a focal esophageal mass is more concerning for an esophageal neoplasm such as adenocarcinoma or squamous cell carcinoma; however, multidetector CT may not allow differentiation between malig-nant and benign esophageal lesions.

Although the diagnosis can often be strongly suggested by a combination of imaging and clini-cal history, further evaluation with upper endos-copy and biopsy is ultimately required for defini-tive diagnosis (Fig 14). Multidetector CT can be useful in differentiating some benign lesions such as fibrovascular polyp and lipoma from esophageal cancer. The former manifests as an intraluminal mass with smooth margins (as compared with

Figure 11. Bronchogenic cyst in a 33-year-old woman. (a) Coned-down axial contrast-enhanced multidetector CT image at the level of the left atrium shows a well-circumscribed homogeneous subcari-nal mass (arrow). The internal Hounsfield unit value was 45, suggesting a soft-tissue mass. (b, c) Coned-down axial T2-weighted (b) and nonenhanced T1-weighted (c) MR images show that the lesion (ar-row) is entirely cystic with a dependent fluid-fluid level (arrowhead). Although many bronchogenic cysts manifest as homogeneous fluid attenuation lesions at multidetector CT, internal heterogeneity may be due to hemorrhagic or proteinaceous com-ponents or infection. In this scenario, MR imaging can be performed to confirm the cystic nature of the abnormality, which demonstrates high signal inten-sity on T2-weighted images regardless of the other contents. Hemorrhagic, proteinaceous, or mucoid content results in variable patterns of signal intensity on T1-weighted images.


Figure 13. Enhancing masses. (a) Paraganglioma in a 29-year-old man. Coned-down axial contrast-enhanced multidetector CT image at the level of the aortic valve shows a heterogeneous mass (M) in the visceral com-partment with internal regions of intense enhancement. The patient presented with clinical symptoms related to catecholamine secretion by the neoplasm, and the catecholamines in blood and urine samples were confir-matory of paraganglioma. (b) Castleman disease in a 42-year-old man. Coned-down axial contrast-enhanced multidetector CT image at the level of the pulmonary trunk shows a large mass (M) in the visceral mediastinum with intense central enhancement and peripheral hypoenhancement. A small right pleural effusion (arrow) is also present. As the patient did not have specific symptoms suggestive of a paraganglioma or a known vascular primary neoplasm, surgical biopsy was performed for diagnostic purposes and revealed Castleman disease.

include composition (soft tissue, fat, calcium, and attenuation), behavior (well-defined or infiltrative), and enhancement characteristics after administration of intravenous contrast material.

In the setting of a known primary malig-nancy, the presence of a cardiac mass is com-patible with metastatic disease until proven otherwise. In the absence of known malignancy, a cardiac mass represents either thrombus or a benign or malignant neoplasm. It is beyond the scope of this article to discuss the distinguishing characteristics of individual cardiac neoplasms;

the irregular margins and lobulated appearance of many esophageal cancers), and both of these benign lesions may demonstrate internal fat attenu-ation due to the presence of adipose tissue (65).

Cardiac Masses.—When a mass involving the heart or pericardium is identified at multi-detector CT, a wide variety of malignant and benign lesions should be considered. One of the most important features is the location of the abnormality, as masses can be categorized as intracavitary, valvular, intramural, or epi-cardial/pericardial. Other significant features

Figure 12. Esophageal duplica-tion cyst in a 33-year-old woman. Coned-down axial contrast-en-hanced multidetector CT image at the level of the aortic arch shows a well-circumscribed, homoge-neous, fluid attenuation mass (M) adjacent to and intimately as-sociated with the upper thoracic esophagus. The location of the ab-normality enables definitive diag-nosis of an esophageal duplication cyst with imaging alone.


however, it is important to emphasize that when these lesions are identified at multidetector CT, further evaluation with more advanced imaging such as electrocardiographically (ECG) gated cardiac CT, cardiac MR imaging, and/or echo-cardiography is necessary for full characteriza-tion (Fig 15).

Approach to the Paravertebral Compartment

General ConsiderationsAs the paravertebral compartment includes the thoracic spine and paravertebral soft tissues, most lesions originating in this region are neoplasms of neurogenic origin. Other less common neoplastic conditions in this compartment include lym-phoma, primary osseous tumors, and metastases. Nonneoplastic causes include thoracic spinal infections due to bacterial and mycobacterial agents, cystic lesions such as thoracic meningo-cele and neurenteric cyst, and extramedullary hematopoiesis.

The approach outlined in this section is based primarily on the composition of the mediastinal abnormality, as there are a limited number of structures from which lesions can arise in the paravertebral compartment. While specific masses

may be included in the differential diagnosis on the basis of imaging features, clinical information is necessary in most cases.

Soft-Tissue Lesions

Neurogenic Neoplasms.—When a smooth, round, or oval mass is present in the paravertebral region at multidetector CT, the most likely diagnosis is a neurogenic neoplasm, typically a benign periph-eral nerve sheath tumor such as schwannoma or neurofibroma. Neurogenic neoplasms, 70%–80% of which are benign, are the most common cause of paravertebral compartment masses and account for 20% and 35% of all adult and pediatric me-diastinal neoplasms, respectively (66). Peripheral nerve sheath neoplasms typically arise from spinal or proximal intercostal nerves, less commonly from the vagus, recurrent laryngeal, or phrenic nerves, and account for 70% of mediastinal neu-rogenic tumors (66). At multidetector CT, periph-

Figure 14. Esophageal cancer in a 49-year-old man. Coned-down axial contrast-enhanced multidetector CT im-age at the level of the left atrium shows marked thickening of the distal thoracic esophagus and an eccentric mass (M) in the right esophageal wall. Upper endoscopy, endoscopic US, and biopsy revealed primary adenocarcinoma of the distal esophagus. The presence of a focal esophageal mass should raise concern for an esophageal neoplasm such as adenocarcinoma or squamous cell carcinoma; however, multidetector CT does not allow differentiation between malignant and benign esophageal lesions, and further im-aging and histologic sampling are ultimately required for definitive diagnosis.

Figure 15. Cardiac angiosarcoma in a 34-year-old man. Coned-down axial contrast-enhanced multidetector CT image shows a large heterogeneous mass (M) arising from the right atrium and extending into the adjacent pericar-dium and mediastinum, compatible with biopsy-proven angiosarcoma. In contrast to the JART model of mediastinal compartments, the heart and pericardium are included in the visceral compartment in the ITMIG model, as are abnor-malities arising from these structures. Once a cardiac mass is identified at multidetector CT, advanced imaging such as ECG-gated cardiac CT, cardiac MR imaging, and/or echo-cardiography is necessary for full characterization.


eral nerve sheath neoplasms may demonstrate a dumbbell morphology and communication with the spinal canal. Regions of heterogeneity may be due to cystic changes or hemorrhage and are more common in schwannomas than in neurofibromas (66) (Fig 16).

Neurogenic neoplasms may cause benign pres-sure erosion of adjacent ribs or vertebrae and en-largement of the neural foramina. These findings are suggestive of a benign lesion, compared with the osseous invasion and destruction typical of malignancies. MR imaging optimally demonstrates the extent of intraspinal/extradural extension, and several imaging signs for peripheral nerve sheath tumors have been described. The “fascicular sign” describes multiple hypointense, small, ring-like structures corresponding to fascicular bundles and is typically associated with schwannomas. The “target sign” is characterized by central low signal intensity and surrounding peripheral high signal intensity and is more commonly seen with neurofi-bromas than with schwannomas.

When a previously stable neurofibroma sud-denly increases in size, develops regions of het-erogeneity, and/or invades adjacent tissues, ma-lignant transformation to a malignant peripheral nerve sheath tumor should be strongly consid-ered (Fig 17). In patients with neurofibromatosis type 1, there is a 10% lifetime risk of developing a malignant peripheral nerve sheath neoplasm, and these lesions account for most cancer-related deaths in these patients. FDG PET/CT has demonstrated utility in distinguishing malignant peripheral nerve sheath tumors from benign neu-rofibromas, with sensitivity of 95% and specificity of 72% (67). Another study revealed sensitivity of 97% and specificity of 87% for detection of malignant peripheral nerve sheath tumors and suggested that lesions with SUVmax less than 2.5 should be considered benign, those with SUVmax

greater than 3.5 should be considered malignant, and those with SUVmax of 2.5–3.5 should be fol-lowed with surveillance imaging (68).

Other neurogenic neoplasms that may originate in the paravertebral compartment include sympa-thetic ganglion neoplasms such as ganglioneuro-mas, ganglioneuroblastomas, and neuroblastomas and neuroendocrine neoplasms such as paragan-gliomas and are much less common. The imag-ing features of other neurogenic neoplasms are often nonspecific, and histologic sampling is often necessary for diagnosis. The presence of intense homogeneous enhancement in a paravertebral mass should raise suspicion for a paraganglioma.

Extramedullary Hematopoiesis.—In the setting of paravertebral masses adjacent to thoracic vertebrae and/or ribs at multidetector CT in a patient with imaging and clinical evidence of a hematologic disorder resulting in bone marrow replacement (myelofibrosis or chronic myelog-enous leukemia) or hemolytic anemia (thalas-semia, sickle cell anemia, or hereditary sphero-cytosis), extramedullary hematopoiesis should be strongly considered (69). These masses may be large or small and unilateral or bilat-eral and typically enhance after administration of intravenous contrast material due to high vascularity. Heterogeneous attenuation and/or enhancement may be seen in the setting of iron deposition and fat infiltration in long-standing lesions (70) (Fig 18).

In the absence of specific osseous findings suggesting an underlying hematologic disorder, functional imaging with 99mTc sulfur colloid bone marrow scanning and SPECT/CT bone marrow scanning may noninvasively confirm the presence of functioning hematopoietic tissue, thus avoiding biopsy. In patients with sickle cell disease, the ancillary finding of autosplenectomy

Figure 16. Neurogenic neoplasm in a 39-year-old asymptomatic woman. Coned-down axial contrast-enhanced multidetector CT image at the level of the aortic arch shows a well-defined heterogeneous soft-tissue mass (arrow) in the right paravertebral medi-astinum, compatible with a benign peripheral nerve sheath tumor, in this case a schwannoma. When a smooth, round, or oval mass is present in the para-vertebral region at multidetector CT, the most likely diagnosis is a neurogenic neoplasm, typically a be-nign peripheral nerve sheath tumor arising from spinal or intercostal nerves. As in this case, internal heterogeneity may be due to cystic changes or hem-orrhage and is more common in schwannomas than in neurofibromas.


Figure 17. Malignant peripheral nerve sheath tumor in a 37-year-old woman with neurofibromatosis type 1. (a) Coned-down axial nonenhanced multidetector CT image of the lower thoracic spine shows a well-circum-scribed homogeneous soft-tissue mass (arrow) in the left paravertebral region, compatible with a biopsy-proven benign peripheral nerve sheath neoplasm. Note the benign pressure erosion exerted on the vertebral body, which further supports the diagnosis. (b) Coned-down axial FDG PET/CT image 2 years later shows marked in-terval increase in the size of the lesion, which demonstrates increased FDG uptake and invasion and destruction of the thoracic spine, left lower lobe, and chest wall—in contrast to the benign pressure erosion seen earlier—findings consistent with malignancy. Findings suggestive of malignant degeneration of a neurofibroma include sudden increase in size, development of internal regions of heterogeneity, invasion of adjacent structures, and increased FDG uptake at PET/CT.

serves as a clue to the diagnosis of extramedul-lary hematopoiesis.

Cystic Lesions

Intrathoracic Meningocele.—When a smooth and unilocular mass with fluid attenuation at multide-tector CT is present in the paravertebral medias-tinum and is associated with vertebral anomalies such as hemivertebrae, butterfly vertebra, or spina bifida, the likely diagnosis is an intrathoracic me-ningocele. An intrathoracic meningocele represents anomalous herniation of the leptomeninges through an intervertebral foramen or vertebral body defect and is more common in adults than in children (56,71). Associated findings include enlargement of intervertebral foramina and vertebral and/or rib anomalies or scoliosis (Fig 19).

Although it may be difficult to distinguish intrathoracic meningoceles from other low-attenuation paravertebral masses such as neurenteric cysts or neurogenic neoplasms, the clinical setting of neurofibromatosis type 1 should clinch the diagnosis. However, in uncertain cases, multidetector CT, MR imag-ing, or myelography performed after intraspinal injection of contrast material will reveal filling of the meningocele (72).

Mediastinal Abscess.—Mediastinal abscess should be considered when a low-attenuation

mass is identified at multidetector CT in a pa-tient after surgery or esophageal perforation or in the setting of infection in the adjacent thorax. Internal foci of air may be present, and commu-nication may be seen with coexisting subphrenic abscesses or empyema (73). In the majority of situations, the clinical context permits definitive diagnosis; however, in selected cases, percuta-neous needle aspiration may be necessary to differentiate between abscess and postoperative seroma or hematoma.

Pancreatic Pseudocyst.—A cystic mass in the paravertebral mediastinum that develops over a short period of time in the clinical setting of pancreatitis may represent intrathoracic extension of a pancreatic pseudocyst (74). These uncom-mon lesions contain pancreatic secretions, blood, and necrotic material and spread through the esophageal or aortic hiatus (69). At multidetector CT, these lesions typically manifest as thin-walled masses that may be isoattenuating or hyperatten-uating depending on the presence of hemorrhage or infection. Separate intra-abdominal pseudo-cysts may or may not be present.

Spinal InfectionsWhen ill-defined soft tissue, unorganized fluid, and/or a loculated collection are present in the paravertebral compartment in a patient with clinical symptoms such as back pain, fever, and


Figure 19. Intrathoracic meningocele in a 49-year-old woman with neurofibromatosis type 1. (a) Coned-down axial contrast-enhanced multidetector CT image at the level of the aortic arch shows a large fluid attenu-ation mass (M) in the posterior right hemithorax that communicates with the spinal canal. Note the irregularity of the adjacent vertebral body and expansion of the spinal canal. (b) Coned-down axial contrast-enhanced multidetector CT image at the level of the kidneys shows multiple cutaneous neurofibromas (arrows). When a homogeneous low-attenuation paravertebral mass is present in a patient with neurofibromatosis type 1 and as-sociated with vertebral anomalies such as hemivertebrae, butterfly vertebra, or spina bifida, a definitive diagnosis of intrathoracic meningocele can be made.

Figure 18. (a) Extramedullary hematopoiesis in a 51-year-old man with myelofibrosis. Coned-down coronal nonenhanced multidetector CT image through the thoracic spine shows numerous paravertebral soft-tissue nodules and masses (arrows), compatible with extramedullary hematopoiesis. (b) Extramedullary hematopoi-esis in a 62-year-old woman with bone marrow replacement due to a chronic myeloid neoplasm. Coned-down axial contrast-enhanced multidetector CT image at the level of the aortic arch shows multiple heterogeneous paravertebral masses (M) composed of both fat and soft tissue, which is characteristic of extramedullary hemato-poiesis. Note the expansion of the marrow spaces of the ribs and vertebral bodies (*), supporting the diagnosis. At multidetector CT, the lesions of extramedullary hematopoiesis may appear as soft-tissue lesions that enhance after administration of intravenous contrast material due to high vascularity or heterogeneous masses due to iron deposition and fat infiltration in long-standing disease.

malaise, spinal infection should be included in the differential diagnosis. These infections are typically due to bacterial organisms, and significant risk fac-tors include diabetes, autoimmune diseases, malig-nancy, immunosuppression, and intravenous drug

use (75). Early multidetector CT findings include paravertebral fat infiltration and intervertebral disk hypoattenuation, whereas osseous erosion, disk space narrowing, and sequestrum formation may be present in later stages (75) (Fig 20).


Figure 20. Discitis/osteomyelitis in a 62-year-old woman. Coned-down axial nonenhanced multidetector CT image (soft-tissue [a] and bone [b] windows) of the lower thoracic spine shows paravertebral soft tissue (arrows in a) and extensive osseous destruction. The clinical presentation was significant for methicillin-resistant Staphy-lococcus aureus (MRSA) bacteremia and back pain. The combination of imaging features and clinical findings enables definitive diagnosis of spinal infection, which may manifest at multidetector CT as ill-defined soft tissue, unorganized fluid, and/or a loculated fluid collection.

In the setting of immunodeficiency, particu-larly human immunodeficiency virus (HIV) infection, tuberculous involvement of the spine should be considered. It has been demonstrated that up to 60% of HIV-positive patients with tuberculosis have skeletal involvement, and the spine is the most commonly affected site, in-volved in approximately 50% of cases. Tubercu-losis associated with involvement of the posterior elements, the presence of a large prevertebral and paravertebral soft-tissue component out of proportion to the degree of osseous destruction, and disk space narrowing enable differentiation of tuberculous from pyogenic infection. Addition-ally, the presence of calcification without new bone formation or sclerosis strongly suggests the diagnosis of Pott disease of the spine (76,77).

ConclusionThe new mediastinal division scheme devel-oped by ITMIG is designed to enable precise identification of mediastinal abnormalities at cross-sectional imaging by radiologists and consistent communication between health care providers. It is anticipated that this system will improve lesion localization, help generate a focused differential diagnosis, and assist in tailoring biopsy and treatment plans. Although mediastinal masses are uncommon, this article presents approaches that radiologists may find helpful when faced with a mediastinal abnor-mality at multidetector CT.

Some mediastinal masses manifest with specific features at multidetector CT that en-

able identification with imaging alone, whereas others may demonstrate suggestive but incon-clusive imaging characteristics. In many cases, a combination of clinical and imaging information permits a presumptive diagnosis. Therefore, the approaches presented here recommend initial inclusion or exclusion of lesions on the basis of multidetector CT features and correlation of less conclusive imaging features with specific clinical information. In many cases, this will strongly suggest a particular diagnosis and a further evaluative or treatment strategy.

Disclosures of Conflicts of Interest.—M.L.R.d.C. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: royalties from Amir-sys Elsevier, Thieme Medical Publishers, and the American Registry of Pathology. Other activities: disclosed no relevant relationships.

references 1. Shields TW. Primary tumors and cysts of the mediastinum.

In: Shields TW, ed. General thoracic surgery. Philadelphia, Pa: Lea & Febiger, 1983; 927–954.

2. Fraser RS, Müller NL, Colman N, Paré PD, eds. The me-diastinum. In: Fraser and Paré’s diagnosis of diseases of the chest. 4th ed. Philadelphia, Pa: Saunders, 1999; 196–234.

3. Fraser RG, Paré PA, eds. The normal chest. In: Diagnosis of diseases of the chest. 2nd ed. Philadelphia, Pa: Saunders, 1977; 1–183.

4. Felson B. Chest roentgenology. Philadelphia, Pa: Saunders, 1973.

5. Heitzman ER. The mediastinum. 2nd ed. New York, NY: Springer-Verlag, 1988.

6. Zylak CJ, Pallie W, Jackson R. Correlative anatomy and computed tomography: a module on the mediastinum. RadioGraphics 1982;2(4):555–592.

7. Whitten CR, Khan S, Munneke GJ, Grubnic S. A diagnos-tic approach to mediastinal abnormalities. RadioGraphics 2007;27(3):657–671.


8. Aquino SL, Duncan G, Taber KH, Sharma A, Hayman LA. Reconciliation of the anatomic, surgical, and radiographic classifications of the mediastinum. J Comput Assist Tomogr 2001;25(3):489–492.

9. Henschke CI, Lee IJ, Wu N, et al. CT screening for lung cancer: prevalence and incidence of mediastinal masses. Radiology 2006;239(2):586–590.

10. Carter BW, Okumura M, Detterbeck FC, Marom EM. Approaching the patient with an anterior mediastinal mass: a guide for radiologists. J Thorac Oncol 2014;9(9 suppl 2):S110–S118.

11. Fujimoto K, Hara M, Tomiyama N, Kusumoto M, Sakai F, Fujii Y. Proposal for a new mediastinal compartment classification of transverse plane images according to the Japanese Association for Research on the Thymus (JART) General Rules for the Study of Mediastinal Tumors. Oncol Rep 2014;31(2):565–572.

12. Carter BW, Tomiyama N, Bhora FY, et al. A modern defini-tion of mediastinal compartments. J Thorac Oncol 2014;9(9 suppl 2):S97–S101.

13. Davis RD Jr, Oldham HN Jr, Sabiston DC Jr. Primary cysts and neoplasms of the mediastinum: recent changes in clinical presentation, methods of diagnosis, management, and results. Ann Thorac Surg 1987;44(3):229–237.

14. Levasseur P, Kaswin R, Rojas-Miranda A, N’guimbous JF, Merlier M, Le Brigand H. Profile of surgical tumors of the mediastinum: apropos of a series of 742 operated patients [in French]. Nouv Presse Med 1976;5(42):2857–2859.

15. Cohen AJ, Thompson L, Edwards FH, Bellamy RF. Primary cysts and tumors of the mediastinum. Ann Thorac Surg 1991;51(3):378–384; discussion 385–386.

16. Rubush JL, Gardner IR, Boyd WC, Ehrenhaft JL. Medias-tinal tumors: review of 186 cases. J Thorac Cardiovasc Surg 1973;65(2):216–222.

17. Wychulis AR, Payne WS, Clagett OT, Woolner LB. Surgi-cal treatment of mediastinal tumors: a 40 year experience. J Thorac Cardiovasc Surg 1971;62(3):379–392.

18. Mullen B, Richardson JD. Primary anterior mediastinal tu-mors in children and adults. Ann Thorac Surg 1986;42(3): 338–345.

19. Takeda S, Miyoshi S, Minami M, et al. Clinical spectrum of primary mediastinal tumors: a comparison of adult and pediatric populations [abstr]. Chest 2000;118:206S.

20. Whooley BP, Urschel JD, Antkowiak JG, Takita H. Primary tumors of the mediastinum. J Surg Oncol 1999;70(2): 95–99.

21. Azarow KS, Pearl RH, Zurcher R, Edwards FH, Cohen AJ. Pri-mary mediastinal masses: a comparison of adult and pediatric populations. J Thorac Cardiovasc Surg 1993;106(1):67–72.

22. Tomiyama N, Honda O, Tsubamoto M, et al. Anterior mediastinal tumors: diagnostic accuracy of CT and MRI. Eur J Radiol 2009;69(2):280–288.

23. Rosado-de-Christenson ML, Templeton PA, Moran CA. Mediastinal germ cell tumors: radiologic and pathologic correlation. RadioGraphics 1992;12(5):1013–1030.

24. Ackman JB, Wu CC. MRI of the thymus. AJR Am J Roent-genol 2011;197(1):W15–W20.

25. Ackman JB. MR imaging of mediastinal masses. Magn Reson Imaging Clin N Am 2015;23(2):141–164.

26. Carter BW, Benveniste MF, Truong MT, Marom EM. State of the art: MR imaging of thymoma. Magn Reson Imaging Clin N Am 2015;23(2):165–177.

27. Inaoka T, Takahashi K, Mineta M, et al. Thymic hyperplasia and thymus gland tumors: differentiation with chemical shift MR imaging. Radiology 2007;243(3):869–876.

28. Takahashi K, Al-Janabi NJ. Computed tomography and magnetic resonance imaging of mediastinal tumors. J Magn Reson Imaging 2010;32(6):1325–1339.

29. Kubota K, Yamada S, Kondo T, et al. PET imaging of primary mediastinal tumours. Br J Cancer 1996;73(7): 882–886.

30. Tatci E, Ozmen O, Dadali Y, et al. The role of FDG PET/CT in evaluation of mediastinal masses and neurogenic tumors of chest wall. Int J Clin Exp Med 2015;8(7): 11146–11152.

31. Luzzi L, Campione A, Gorla A, et al. Role of fluorine-fluorodeoxyglucose positron emission tomography/com-puted tomography in preoperative assessment of anterior

mediastinal masses. Eur J Cardiothorac Surg 2009;36(3): 475–479.

32. Sung YM, Lee KS, Kim BT, Choi JY, Shim YM, Yi CA. 18F-FDG PET/CT of thymic epithelial tumors: usefulness for distinguishing and staging tumor subgroups. J Nucl Med 2006;47(10):1628–1634.

33. Treglia G, Sadeghi R, Giovanella L, Cafarotti S, Filosso P, Lococo F. Is (18)F-FDG PET useful in predicting the WHO grade of malignancy in thymic epithelial tumors? A meta-analysis. Lung Cancer 2014;86(1):5–13.

34. Jerushalmi J, Frenkel A, Bar-Shalom R, Khoury J, Israel O. Physiologic thymic uptake of 18F-FDG in children and young adults: a PET/CT evaluation of incidence, patterns, and relationship to treatment. J Nucl Med 2009;50(6):849–853.

35. Detterbeck F. Clinical approach to mediastinal masses. In: Kuzdal JML, Muller M, Papagiannopoulos K, et al, eds. ESTS textbook of thoracic surgery. Exeter, England: Euro-pean Society of Thoracic Surgeons, 2014.

36. Naidich DP, Webb WR, Muller NL, Krinsky GA, Zerhouni EA, Siegelman SS, eds. Mediastinum. In: Computed tomog-raphy and magnetic resonance of the thorax. 3rd ed. Phila-delphia, Pa: Lippincott Williams & Wilkins, 1999;82–83.

37. Molinari F, Bankier AA, Eisenberg RL. Fat-containing lesions in adult thoracic imaging. AJR Am J Roentgenol 2011;197(5):W795–W813.

38. Wright C. Germ cell tumors of the mediastinum. In: Pearson F, Cooper J, Deslauriers J, et al, eds. Thoracic surgery. New York, NY: Churchill Livingstone, 2002; 1711–1719.

39. Gaerte SC, Meyer CA, Winer-Muram HT, Tarver RD, Conces DJ Jr. Fat-containing lesions of the chest. Radio-Graphics 2002;22(Spec Issue):S61–S78.

40. Nishino M, Ashiku SK, Kocher ON, Thurer RL, Boiselle PM, Hatabu H. The thymus: a comprehensive review. Ra-dioGraphics 2006;26(2):335–348.

41. Munden RF, Nesbitt JC, Kemp BL, Chasen MH, Whitman GJ. Primary liposarcoma of the mediastinum. AJR Am J Roentgenol 2000;175(5):1340.

42. Hahn HP, Fletcher CD. Primary mediastinal liposarcoma: clinicopathologic analysis of 24 cases. Am J Surg Pathol 2007;31(12):1868–1874.

43. Nasseri F, Eftekhari F. Clinical and radiologic review of the normal and abnormal thymus: pearls and pitfalls. Radio-Graphics 2010;30(2):413–428.

44. Jeung MY, Gasser B, Gangi A, et al. Imaging of cystic masses of the mediastinum. RadioGraphics 2002;22(Spec Issue):S79–S93.

45. Kissin CM, Husband JE, Nicholas D, Eversman W. Benign thymic enlargement in adults after chemotherapy: CT dem-onstration. Radiology 1987;163(1):67–70.

46. Benveniste MF, Rosado-de-Christenson ML, Sabloff BS, Moran CA, Swisher SG, Marom EM. Role of imaging in the diagnosis, staging, and treatment of thymoma. RadioGraphics 2011;31(7):1847–1861; discussion 1861–1863.

47. Priola AM, Priola SM, Di Franco M, Cataldi A, Durando S, Fava C. Computed tomography and thymoma: distinc-tive findings in invasive and noninvasive thymoma and predictive features of recurrence. Radiol Med (Torino) 2010;115(1):1–21.

48. Tomiyama N, Müller NL, Ellis SJ, et al. Invasive and non-invasive thymoma: distinctive CT features. J Comput Assist Tomogr 2001;25(3):388–393.

49. Schaefer NG, Hany TF, Taverna C, et al. Non-Hodgkin lymphoma and Hodgkin disease: coregistered FDG PET and CT at staging and restaging—do we need contrast-enhanced CT? Radiology 2004;232(3):823–829.

50. Strollo DC, Rosado-de-Christenson ML. Primary mediastinal malignant germ cell neoplasms: imaging features. Chest Surg Clin N Am 2002;12(4):645–658.

51. Lemarié E, Assouline PS, Diot P, et al. Primary mediastinal germ cell tumors: results of a French retrospective study. Chest 1992;102(5):1477–1483.

52. Economou JS, Trump DL, Holmes EC, Eggleston JE. Man-agement of primary germ cell tumors of the mediastinum. J Thorac Cardiovasc Surg 1982;83(5):643–649.

53. Lee KS, Im JG, Han CH, Han MC, Kim CW, Kim WS. Malignant primary germ cell tumors of the mediastinum: CT features. AJR Am J Roentgenol 1989;153(5):947–951.


54. Wright CD, Kesler KA, Nichols CR, et al. Primary me-diastinal nonseminomatous germ cell tumors: results of a multimodality approach. J Thorac Cardiovasc Surg 1990;99 (2):210–217.

55. Kesler KA, Rieger KM, Ganjoo KN, et al. Primary medi-astinal nonseminomatous germ cell tumors: the influence of postchemotherapy pathology on long-term survival after surgery. J Thorac Cardiovasc Surg 1999;118(4): 692–700.

56. Juanpere S, Cañete N, Ortuño P, Martínez S, Sanchez G, Bernado L. A diagnostic approach to the mediastinal masses. Insights Imaging 2013;4(1):29–52.

57. Rodgers SE, Hunter GJ, Hamberg LM, et al. Improved preop-erative planning for directed parathyroidectomy with 4-dimen-sional computed tomography. Surgery 2006;140(6):932–940; discussion 940–941.

58. Kukar M, Platz TA, Schaffner TJ, et al. The use of modi-fied four-dimensional computed tomography in patients with primary hyperparathyroidism: an argument for the abandonment of routine sestamibi single-positron emission computed tomography (SPECT). Ann Surg Oncol 2015;22 (1):139–145.

59. Bahl M, Muzaffar M, Vij G, Sosa JA, Choudhury KR, Hoang JK. Prevalence of the polar vessel sign in parathyroid adenomas on the arterial phase of 4D CT. AJNR Am J Neuroradiol 2014;35(3):578–581.

60. Ferguson CC, Young LN, Sutherland JB, Macpherson RI. Intrathoracic gastrogenic cyst: preoperative diagnosis by technetium pertechnetate scan. J Pediatr Surg 1973;8(5): 827–828.

61. Balcombe J, Torigian DA, Kim W, Miller WT Jr. Cross-sectional imaging of paragangliomas of the aortic body and other thoracic branchiomeric paraganglia. AJR Am J Roent-genol 2007;188(4):1054–1058.

62. Brown ML, Zayas GE, Abel MD, Young WF Jr, Schaff HV. Mediastinal paragangliomas: the Mayo Clinic experience. Ann Thorac Surg 2008;86(3):946–951.

63. Bonekamp D, Horton KM, Hruban RH, Fishman EK. Castleman disease: the great mimic. RadioGraphics 2011;31(6):1793–1807.

64. McAdams HP, Rosado-de-Christenson M, Fishback NF, Templeton PA. Castleman disease of the thorax: radiologic features with clinical and histopathologic correlation. Radiol-ogy 1998;209(1):221–228.

65. Lewis RB, Mehrotra AK, Rodriguez P, Levine MS. Esophageal neoplasms: radiologic-pathologic correlation. RadioGraphics 2013;33(4):1083–1108.

66. Strollo DC, Rosado-de-Christenson ML, Jett JR. Primary mediastinal tumors. II. Tumors of the middle and posterior mediastinum. Chest 1997;112(5):1344–1357.

67. Bredella MA, Torriani M, Hornicek F, et al. Value of PET in the assessment of patients with neurofibromatosis type 1. AJR Am J Roentgenol 2007;189(4):928–935.

68. Warbey VS, Ferner RE, Dunn JT, Calonje E, O’Doherty MJ. [18F]FDG PET/CT in the diagnosis of malignant peripheral nerve sheath tumours in neurofibromatosis type-1. Eur J Nucl Med Mol Imaging 2009;36(5):751–757.

69. Berkmen YM, Zalta BA. Case 126: extramedullary hemato-poiesis. Radiology 2007;245(3):905–908.

70. Georgiades CS, Neyman EG, Francis IR, Sneider MB, Fishman EK. Typical and atypical presentations of extra-medullary hemopoiesis. AJR Am J Roentgenol 2002;179(5): 1239–1243.

71. Miles J, Pennybacker J, Sheldon P. Intrathoracic meningocele: its development and association with neurofibromatosis. J Neurol Neurosurg Psychiatry 1969;32(2):99–110.

72. Webb WR. Diseases of the mediastinum. In: Putman CE, Ravin CE, eds. Textbook of diagnostic imaging. 2nd ed. Philadelphia, Pa: Saunders, 1994; 428–447.

73. Glazer HS, Siegel MJ, Sagel SS. Low-attenuation medias-tinal masses on CT. AJR Am J Roentgenol 1989;152(6): 1173–1177.

74. Kirchner SG, Heller RM, Smith CW. Pancreatic pseudocyst of the mediastinum. Radiology 1977;123(1):37–42.

75. Meyer CA, Vagal AS, Seaman D. Put your back into it: pathologic conditions of the spine at chest CT. RadioGraphics 2011;31(5):1425–1441.

76. De Backer AI, Mortelé KJ, Vanschoubroeck IJ, et al. Tuber-culosis of the spine: CT and MR imaging features. JBR-BTR 2005;88(2):92–97.

77. Engin G, Acunaş B, Acunaş G, Tunaci M. Imaging of extrapul-monary tuberculosis. RadioGraphics 2000;20(2):471–488; quiz 529–530, 532.

This journal-based SA-CME activity has been approved for AMA PRA Category 1 CreditTM. See www.rsna.org/education/search/RG.

413 itmig classification of mediastinal compartments and ... › ... › 2017 › 2_mayo › rsna...

Documents