Ann Thorac Surg 2003;75:1086-1090
© 2003 The Society of Thoracic Surgeons
Original article: general thoracic
The value of a noninvasive diagnostic approach to mediastinal masses
Ruediger Hoerbelt, MDa*,
Lars Keuneckea,
Helmuth Grimm, MDa,
Konrad Schwemmle, MDa,
Winfried Padberg, MDa
a Department of General and Thoracic Surgery, Justus-Liebig-University of Gießen, Gießen, Germany
Accepted for publication November 1, 2002.
* Address reprint requests to Dr Hoerbelt, Transplantation Biology Research Center, Massachusetts General Hospital, Bldg 149, 13th St, Charlestown, MA 02129, USA
e-mail: ruediger.hoerbelt{at}tbrc.mgh.harvard.edu
 |
Abstract
|
|---|
BACKGROUND: Mediastinal tumors show a wide variability, and therefore, a standardized diagnostic and therapeutic workup is instrumental. We subdivided mediastinal tumors into nonlymphatic mediastinal tumors (NLMTs), most of which require surgical resection without need of preoperative histology, and mediastinal lymphadenopathy (MLA), requiring surgical biopsy for exact histologic classification. We investigated the accuracy of noninvasive diagnostic studies distinguishing between the two groups of MLA and NLMT.
METHODS: A retrospective analysis was performed on patients who had previously undergone surgery on mediastinal tumors. Their data were statistically analyzed (
2 test, logistic regression analysis), and the values of medical history, physical examination, laboratory tests, and computerized tomography scan discriminating between MLA and NLMT were assessed.
RESULTS: Out of 299 patients included in the study, 242 (80.9%) had MLA and 57 (19.1%) had NLMT. Sensitivity and specificity of noninvasive investigations for differentiation of MLA and NLMT were 98.2% and 86.0%, respectively. Whereas the prevalence of thoracic symptoms such as shortness of breath, cough, or chest pain was similar in both groups (MLA, 165 [69.3%]; NLMT, 41 [69.5%]; p = 0.98), systemic symptoms, including fever, night sweats, or weight loss (MLA, 110 [49.8%]; NLMT, 17 [29.3%]; p < 0.01), and signs of inflammation, such as c-reactive protein, erythrocyte sedimentation rate, and leukocytosis (MLA, 202 [85.6%]; NLMT, 34 [57.6%]; p < 0.001), were significantly more common in MLA.
CONCLUSIONS: Noninvasive diagnostic procedures, including medical history, physical examination, laboratory tests, and computerized tomographic scan, are highly sensitive in detecting MLAs that should undergo surgical biopsy. Our data suggest confirming all suspected NLMTs by fine needle aspiration (FNA) biopsy before surgery.
|
Introduction
|
|---|
Mediastinal tumors comprise various benign and malignant neoplasms that share the same anatomic location within the thorax. Histologic classification of mediastinal tumors show a wide variability, and 25% to 49% of these masses are malignant [1, 2]. Therefore, a standardized diagnostic and therapeutic regimen, not delayed by unnecessary diagnostic steps, is instrumental.
Once a mediastinal mass has been detected (usually by standard chest radiograph), diagnostic evaluation includes medical history, physical examination, laboratory tests, and computerized tomographic (CT) scan. The latter provides information about anatomic location, extent of disease, tissue invasion, and tissue density [3, 4]. Despite great variability, further clinical management for mediastinal tumors is relatively uniform, in that patients generally require further invasive diagnostic evaluation [3]. Common techniques to obtain the histology of mediastinal tumors include image-guided fine needle aspiration biopsy, surgical biopsy through various approaches, and complete resection. The clinically important aim of noninvasive diagnostic investigations is to guide the choice of invasive approach. To simplify this process, mediastinal tumors can be subdivided into mediastinal lymphadenopathy (MLA) and nonlymphatic mediastinal tumors (NLMTs). Differential diagnosis of MLA ranges from lymph node metastases, malignant lymphomas, infectious, immunologic, endocrine, and lipid storage diseases to disorders such as sarcoidosis, histiocytosis X, and Castleman’s disease [5]. Treatment for this heterogeneous group of diseases is primarily nonsurgical, and exact histologic classification has great impact on the therapeutic approach [6]. Because relatively large tissue samples are required for definite diagnosis of MLA, image-guided fine needle aspiration (FNA) biopsies are often insufficient [7] and surgical biopsy should be performed. Most nonlymphatic mediastinal tumors, even when benign, require surgical resection, and for many of these lesions, attempting preoperative invasive studies is unnecessary and potentially risky [3].
The aim of this study was to determine how accurate noninvasive diagnostic studies, including medical history, physical examination, laboratory tests, and CT scan, help distinguish between MLAs requiring surgical biopsy and NMLTs, most of which necessitate surgical resection.
|
Material and methods
|
|---|
All records from our clinic between January 1986 and July 1999 were reviewed to identify patients with mediastinal tumors. Patients were only eligible if histology had been established and sufficient data about medical history, physical examination, and diagnostic findings were obtainable. Patients with esophageal malignancies and hernias into the mediastinum were excluded. Information about the patients was obtained through surgery clinic charts, in-patient records, pathology records, operative records, and hematology oncology clinic records.
Each chart was reviewed for demographic data, clinical presentation, and diagnostic investigation. The operative approach and procedure, the location, and the histologic diagnosis of the tumor were recorded.
For the purposes of our study, we divide the mediastinum into anterior, middle, and posterior mediastinum, as described by Burkell and associates [8].
The different histologic entities were allocated to one of two groups. All types of inflammatory, specific, and nonspecific infectious lymphatic lesions, metastases, and malignant lymphomas were subsumed into the MLA group. The NLMT group included thymic, cystic, germ cell, vascular, neurogenic, and endocrine tumors, as well as mediastinal goiters, lymphangiomas, and carcinomas.
Sensitivity and specificity were assessed for the complete noninvasive diagnostic approach, consisting of medical history, physical and laboratory findings, chest roentgenogram, and CT scan. The histomorphological diagnosis was taken as the gold standard. Sensitivity was defined as the number of MLAs with positive results as a percentage of the total number of MLAs. Specificity was defined as the number of NLMTs with negative results as a percentage of the total number of NLMTs. Cases with unclear results after noninvasive diagnostic workup were excluded from the calculation of sensitivity and specificity.
Logistic regression analysis was used to investigate the role of systemic symptoms and signs of inflammation as predictors of MLA. We defined fever, night sweats, and weight loss (more than 10% in 6 months) as systemic symptoms. We generalized chest pain and respiratory symptoms such as cough, hoarseness, and shortness of breath under "thoracic symptoms." As signs of inflammation we investigated the white blood cell count (WBC), the erythrocyte sedimentation rate (ESR), and the C-reactive protein (CRP). These were defined as elevated, if WBC was greater than 10 x 109/L, ESR was higher than 30 mm/h, and CRP was greater than 5 mg/dL. Statistical analyses were performed on all signs and symptoms using the 2 test, and p less than 0.05 was accepted as significant. SPSS (version 9.0; SSPS Inc., Chicago, IL) was used to calculate sensitivity and specificity and to conduct the statistical test.
|
Results
|
|---|
Between January 1986 and July 1999, 354 patients with mediastinal masses underwent operations in our clinic. After exclusion of esophageal malignancies, cardiac tumors, and hernias, there were 323 patients remaining in our database. Twenty-four patients had to be excluded because of insufficient histologic or diagnostic data. Thus, there were 299 patients included in the study.
Tumor histology and anatomical location
The distributions of histologic entities and of tumor locations within each group are shown in Table 1.
Following the categorization system described above, we subdivided the whole series into groups of MLAs and NLMTs. The majority of tumors (242/299; 80.9%) were MLAs, most of which were metastases (135/242; 45.2%). Both categories of tumors were most commonly located in the anterior mediastinum (86.4% and 80.7%, respectively).
Demographic data
There were 91 female and 208 male patients in the study. The ratio of female to male patients was 1:2.81 for MLA and 1:1.11 for NLMT (p = 0.002). For the whole series, mean age was 53.14 ± 15.1 years (range, 15 to 84 years). The mean age was 53.48 ± 13.87 years in the MLA group and 51.78 ± 19.39 years in the NLMT group. In both groups, the majority of patients, 179 (73.9%) of MLA and 31 (54.4%) of NLMT, were between 41 and 70 years of age. The number of patients younger than 20 years was 3 (1.0%) in the MLA group (all of them had specific infections) and 7 (12.2%) in the NLMT group (three germ cell and four neurogenic tumors).
Surgical approach and procedures
The surgical approach to our patients included diagnostic mediastinoscopy with biopsy in 230 (76.9%) patients, thoracotomy in 38 (12.7%) patients, video-assisted thoracoscopy in 1 (0.3%) patient, median sternotomy in 11 (3.7%) patients, and cervical incisions (for mediastinal goiters) in 12 (4.0%) patients. Seven (2.3%) patients had parasternal mediastinotomy with biopsy for retrosternal mediastinal tumors.
Mediastinal lymphadenopathy
Two hundred thirty-four (97.5%) patients with MLA underwent diagnostic procedures to obtain the histology. Six (2.5%) patients who were preoperatively diagnosed with thymoma or mediastinal cysts underwent complete tumor excision by sternotomy (2 patients) or thoracotomy (4 patients).
Nonlymphatic mediastinal tumors
In the NLMT group, 48 (81.4%) patients received a curative tumor excision. Two patients with elevated levels of human chorionic gonadotropin and 
-fetoprotein were biopsied by mediastinoscopy and subsequently treated with chemotherapy followed by surgical resection. In 7 patients, the mediastinal tumor infiltrated the great mediastinal vessels and biopsies were taken by mediastinoscopy. Two thymus tumors had been mistaken for lymphomas after noninvasive workup, and mediastinoscopy was performed for biopsy.
|
Symptoms and signs
|
|---|
Symptoms and signs in our series are shown in Table 2.
There were 243 (81.3%) symptomatic patients in our series. The symptomatic population of either group, MLA and NLMT, was 81.2% and 81.4%, respectively. Common symptoms, evenly distributed between both groups, were thoracic symptoms such as shortness of breath, cough, and chest pain.
Systemic symptoms such as weight loss, night sweats, or fever (p = 0.005) and laboratory signs of inflammation such as C-reactive protein, leukocytosis, and erythrocyte sedimentation rate (p < 0.001) were significantly more common in patients with mediastinal lymphadenopathy, as compared with nonlymphatic mediastinal tumors. To exclude bias caused by the great number of infectious diseases within the MLA group, we compared the presence of inflammatory signs in benign and malignant MLAs and found no significant difference (76.1% in benign MLAs vs 89% in malignant MLAs; p = 0.1). Presence of systemic symptoms and signs of inflammation were highly predictive of lymphatic mediastinal tumors (p = 0.001), as demonstrated by logistic regression analysis.
|
Diagnostic accuracy of noninvasive diagnostic approach
|
|---|
The noninvasive diagnostic approach included medical history, physical findings, laboratory results, chest radiograph, and CT scan. All patients were initially evaluated with posteroanterior and lateral chest radiograph. CT scan was used in 281 (94.0%) patients.
|
Tumors of uncertain histology
|
|---|
In 35 (11.7%) patients, the entity of the mediastinal mass could not be defined, as to allocate them to the MLA or NLMT groups. Twenty-six of these tumors were histologically diagnosed as MLA: four malignant lymphomas, three sarcoidosis, two lymph node tuberculosis, one nonspecific infectious lymphadenopathy, and 16 metastases. In nine of the unclear mediastinal masses, the histologic evaluation revealed NLMT: three thymic tumors, three mesenchymal tumors, two neurogenic tumors, and one germ cell tumor.
|
Mediastinal lymphadenopathy
|
|---|
In the MLA group, 216 (90.0%) patients were correctly identified. Four (1.7%) MLAs (one Hodgkin’s, three nonspecific lymphadenopathy) were mistakenly diagnosed as NLMT.
|
Nonlymphatic mediastinal tumors
|
|---|
Among the patients with NLMT, 37 (67.3%) were correctly diagnosed by noninvasive investigations. In six cases (10.5%), NLMT was confused with LMT (two thymomas, one thymic tumor, one carcinoid, one teratoma, one cystic lymphangioma). The statistical analysis for the noninvasive diagnostic approach is shown in Table 3.
View this table:
[in this window]
[in a new window]
|
Table 3. Discrimination Between MLA and NLMTs by Noninvasive Diagnostic Workup
MLA = mediastinal lymphadenopathy; NLMT = nonlymphatic mediastinal tumor.
|
|
|
Comment
|
|---|
Given the wide variability of histology presenting as mediastinal mass, a standardized diagnostic and therapeutic workup is instrumental. In our study, we investigated the accuracy of noninvasive diagnostic studies, including medical history, laboratory tests, chest radiograph, and CT scan, representing the common first-line diagnostic approach to mediastinal masses [3]. We addressed the question of whether this basic diagnostic approach is sufficient in separating lymphatic tumors that require tumor biopsy before therapeutic management from NLMTs that in general necessitate complete surgical resection without need for preoperative biopsy.
In our series, the vast majority of tumors were lymphatic (240; 82.2%), with sarcoidosis being the most common finding. In 59 (19.8%) patients, a nonlymphatic mediastinal tumor was diagnosed, and thyroid adenomas were most frequent in this group. The great proportion of lymphatic lesions differs from other large series [9–11], but is not surprising considering our inclusion criteria. Because we were interested in diagnostic quality, we included all kinds of tumors that could challenge the diagnostic algorithm.
Our results provide good evidence that noninvasive diagnostic studies allow detection of lymphatic mediastinal tumors with high sensitivity (98.2%). After diagnosis, these tumors need to be adequately biopsied in order to establish histology. Definitive diagnosis of lymphatic tumors requires relatively large tissue samples, and those obtained by percutaneous image-guided FNA biopsies are often insufficient [7, 12]. Because noninvasive studies obviously can distinguish LMTs with high sensitivity, and given the low accuracy of FNA biopsies for lymphatic tumors, we propose that patients in whom noninvasive studies are suspicious for MLA should undergo surgical biopsy either by mediastinoscopy [13] or mediastinotomy. As opposed to sensitivity, our results show relatively low specificity of noninvasive studies for mediastinal lymphadenopathy (86%). Accordingly, patients in whom noninvasive diagnostic workup revealed a nonlymphatic lesion have a considerable chance of being false negative for MLA. These data suggest that one should generally confirm nonlymphatic mediastinal tumors by performing a preoperative biopsy. Because FNA biopsy has been shown to be highly accurate for nonlymphatic lesions [12, 14], this less-invasive approach appears to be appropriate in NLMT. In 12.1% of cases, noninvasive diagnostic investigations were not sufficient, qualifying the tumor as either lymphatic or nonlymphatic. This unexpectedly high number of tumors with indeterminate classification may be explained by the fact that radiologists in our study did not particularly focus on discrimination between MLA and NLMT.
As an attempt to improve the diagnostic accuracy of noninvasive diagnoses, we looked at predictive factors for mediastinal lymphadenopathy. Our results show that more than 80% of patients became symptomatic during the clinical course of the mediastinal tumor, which is consistent with reports by others [15]. The majority of symptoms were thoracic symptoms (cough, chest pain, or shortness of breath), and there was no significant difference between MLA and NLMT (p = 0.98). However, if we looked at occurrence of systemic symptoms (p = 0.005) and systemic signs of inflammation (p < 0.001), we found a significantly higher rate in patients with MLA (see Table 2). The most likely explanation for the significant finding is that most kinds of lymphadenopathy are an expression of systemic disease leading to systemic effects. One might argue that the great proportion of infectious lesions in our series could explain the high rate of inflammatory signs. However, the comparison of inflammatory signs in benign and malignant MLA did not reveal any significant difference.
We conclude that the basic diagnostic workup, including medical history, physical and laboratory examination, and CT scan, can reliably detect mediastinal lymphadenopathy, and consideration of inflammatory signs may increase the diagnostic accuracy. These patients should directly undergo surgical biopsy in order to confirm and further refine the diagnosis. According to the relatively low specificity found in our study, the diagnosis of NLMT should be confirmed by fine-needle aspiration biopsy in order to avoid unnecessary surgical resections.
|
Acknowledgments
|
|---|
We thank Dr Douglas R. Johnston for critically reading the manuscript.
|
References
|
|---|
- Temes R., Chavez T., Mapel D., et al. Primary mediastinal malignancies: findings in 219 patients. West J Med 1999;170:161-166.[Medline]
- Cohen A.J., Thompson L., Edwards F.H., Bellamy R.F. Primary cysts and tumors of the mediastinum. Ann Thorac Surg 1991;51:378-384.[Abstract]
- Yoneda K.Y., Louie S., Shelton D.K. Mediastinal tumors. Curr Opin Pulmon Med 2001;7:226-233.[Medline]
- Kohman L.J. Approach to the diagnosis and staging of mediastinal masses. Chest 1993;103(Suppl):328-330.[Free Full Text]
- Henry PH, Longo DL. Enlargement of lymph nodes and spleen. In Fauci AS, Braunwald E, Isselbacher KJ, et al, eds. Harrison’s principles of internal medicine. New York, NY: McGraw-Hill Companies, 1998:346
- Bacha E.A., Chapelier A.R., Macchiarini P., Fadel E., Dartevelle P.G. Surgery for invasive primary mediastinal tumors. Ann Thorac Surg 1998;66:234-239.[Abstract/Free Full Text]
- Robinson L.A., Dobson J.R., Bierman P.J. Fallibility of transthoracic needle biopsy of anterior mediastinal masses. Thorax 1995;50:1114-1116.[Abstract/Free Full Text]
- Burkell CC, Cross JM, Kent HP, Nanson EM. Mass lesions of the mediastinum. Curr Probl Surg 1969:2–57
- Benjamin S.P., McCormack L.J., Effler D.B., Groves L.K. Primary tumors of the mediastinum. Chest 1972;62:297-303.[Abstract/Free Full Text]
- Parish J.M., Rosenow E.C., III, Muhm J.R. Mediastinal masses: clues to interpretation of radiologic studies. Postgrad Med 1984;76:173-176.
- Sabiston D.C., Jr, Blalock A., Scott H.W. Primary neoplasms and cysts of the mediastinum. Ann Surg 1952;136:777-797.[Medline]
- Morrissey B., Adams H., Gibbs A.R., Crane M.D. Percutaneous needle biopsy of the mediastinum: review of 94 procedures. Thorax 1993;48:632-637.[Abstract/Free Full Text]
- Porte H., Roumilhac D., Eraldi L., Cordonnier C., Puech P., Wurtz A. The role of mediastinoscopy in the diagnosis of mediastinal lymphadenopathy. Eur J Cardiothorac Surg 1998;13:196-199.
- Hsu W.H., Chiang C.D., Hsu J.Y., Kwan P.C., Chen C.L., Chen C.Y. Ultrasonically guided needle biopsy of anterior mediastinal masses: comparison of carcinomatous and non-carcinomatous masses. J Clin Ultrasound 1995;23:349-356.[Medline]
- Wongsangiem M., Tangthangtham A. Primary tumors of the mediastinum: 190 cases analysis (1975–1995). J Med Assoc Thai 1996;79:689-697.[Medline]
This article has been cited by other articles:
|
|
|
|
 
P. R. Koduri
The Diagnostic Approach to Mediastinal Masses
Ann. Thorac. Surg.,
November 1, 2004;
78(5):
1888 - 1888.
[Full Text]
[PDF]
|
|
|
|
|
|
|
R. Hoerbelt and W. Padberg
Reply
Ann. Thorac. Surg.,
November 1, 2004;
78(5):
1888 - 1888.
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. Sami Bayram, B. Ozdemir, F. Aydiner, and S. Gullulu
Carcinosarcoma as a primary mediastinal tumor
Interactive CardioVascular and Thoracic Surgery,
March 1, 2004;
3(1):
11 - 13.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
