MRI and CT in the Differential Diagnosis of Pleural Disease

MRI and CT in the Differential Diagnosis of Pleural Disease

MRI and CT in the Differential Diagnosis of Pleural Disease* Johannes Hierholzer, MD; Liangping Luo, MD; Roland C. Bittner, MD; Christian Stroszczynsk...

1MB Sizes 2 Downloads 101 Views

MRI and CT in the Differential Diagnosis of Pleural Disease* Johannes Hierholzer, MD; Liangping Luo, MD; Roland C. Bittner, MD; Christian Stroszczynski, MD; Ralf-Ju¨rgen Schro¨der, MD; Nicolas Schoenfeld, MD; Peter Dorow, MD; Robert Loddenkemper, MD, FCCP; and Anne Grassot, MD

Study objective: To explore the role of MRI in the differential diagnosis of pleural disease. Patients: Forty-two patients with pleural disease were included. Method: Retrospective study. All patients were examined with both CT and MRI. The morphologic features of pleural lesions and magnetic resonance signal intensity on T1-weighted, T2-weighted, and contrast-enhanced T1-weighted images were evaluated. Results: Mediastinal pleural involvement, circumferential pleural thickening, nodularity, irregularity of pleural contour, and infiltration of the chest wall and/or diaphragm were most suggestive of a malignant cause both on CT and MRI. Pleural calcification on CT was suggestive of a benign cause. Contrary to what has been previously reported in the literature, neither on CT nor on MRI, pleural thickness > 1 cm revealed significant difference between malignant and benign pleural disease (p > 0.05, ␹2 test). High signal intensity in relation to intercostal muscles on T2-weighted and/or contrast-enhanced T1-weighted images was significantly suggestive for a malignant disease. Using morphologic features in combination with the signal intensity features, MRI had a sensitivity of 100% and a specificity of 93% in the detection of pleural malignancy. Conclusion: When signal intensity and morphologic features are assessed, MRI is more useful and therefore superior to CT in differentiation of malignant from benign pleural disease. (CHEST 2000; 118:604 – 609) Key words: CT; MRI; pleural disease; signal intensity Abbreviation: HU ⫽ Hounsfield units

number of different imaging modalities may be A used in the diagnosis and management of pleural disease. The basic investigative tool is the chest radiograph, which may be diagnostic. But depending on the clinical context, the optimal technique for further evaluation may be sonography, CT, or MRI. Furthermore, the imaging differentiation of benign and malignant pleural disease is important because the specific diagnosis is often difficult to make by clinical criteria, pleurocentesis, or percutaneous pleural biopsy.1 Therefore, the aim of this study was *From the Klinik und Poliklinik fu¨r Strahlenheilkunde (Drs. Hierholzer, Bittner, Stroszczynski, and Schro¨der), Charite´ Virchow-Klinikum, Humboldt Universitt¨ zu Berlin, Germany; the Medical Imaging Center (Dr. Luo), The First Affiliated Hospital, Jinan University Medical College, Guangzhou, People’s Republic of China; Lungenklinik Heckeshorn (Drs. Schoenfeld, Loddenkemper, and Grassot), Berlin, Germany; and DRK-Krankenhaus (Dr. Dorow), Drontheimer Strasse, Berlin, Germany. Manuscript received September 21, 1999; revision accepted March 30, 2000. Correspondence to: Johannes Hierholzer, MD, Radiologische Klinik, Klinikum Ernst-von-Bergmann, Charlottenstrasse 72, 14467 Potsdam, Germany 604

to compare MRI and CT in the separation of benign from malignant pleural disease.

Materials and Methods Between January 1992 and June 1998, 88 consecutive patients were referred to the Charite´, University Hospital of the Humboldt University in Berlin, Germany, for MRI of pleural disease. The following criteria were chosen to include the patients into our study: (1) histologic confidence of diagnosis, (2) examination by both CT and MRI at least once, and (3) CT findings of examination performed within 4 weeks of MRI available, and MRI performed within 2 weeks of established histologic diagnosis. Forty-six patients did not match the entry criteria due to missing final histologic diagnosis (18 patients with high probability of benign disease), prolonged CT-MRI time interval (23 patients), and delayed surgical confirmation (5 patients). Therefore, 42 of 88 patients matched these inclusion criteria and were included in this study. Twenty-six patients were men and 16 were women, with a mean age of 55.9 years (range, 22 to 82 years). The final diagnosis of all patients was established by thoracoscopy (n ⫽ 16), thoracotomy (n ⫽ 10), pleural fluid cytology (n ⫽ 8), and repeated percutaneous biopsy (n ⫽ 8). Final histologic diagnosis revealed 27 cases of malignant and 15 cases of benign pleural disease (Table 1). Clinical Investigations

Table 1—Final Diagnoses of Pleural Lesions Examined Type of Lesion Malignant (n ⫽ 27) Malignant mesothelioma Pleural metastases Chest wall invasion of bronchogenic carcinoma Pleural liposarcoma Malignant fibrous histiocytoma Benign (n ⫽ 15) Diffuse pleural fibrosis Tuberculous pleurisy Asbestos pleurisy Localized benign fibroma Pleural fibromatosis Pleural hyalinosis

No. of Patients 9 12 4 1 1 5 5 2 1 1 1

All CT scans were obtained on one of three CT scanners (Somatom DRH/Plus/Plus S; Siemens; Erlangen, Germany). Contiguous 5/8/10-mm collimation scans were obtained from lung apices to the level of the adrenals. All scans were photographed at window settings appropriate for lung parenchyma (window level, ⫺ 600 to 700 Hounsfield units [HU]; width, 1,300 to 1,600 HU) and mediastinum (window level, 30 to 50 HU; width, 350 to 500 HU). Thirty-three patients were examined with a contrast-enhanced CT after a bolus administration of 100 to 120 mL contrast material (Ultravist; Schering AG; Berlin, Germany). All CT examinations were performed in other institutions prior to referred to our hospital. MRI was acquired on a 1.5-T magnetic resonance system (Magnetom SP 4000; Siemens). Using a body coil, a heart ratedependent T1-weighted sequence (500 to 900/10 ms), with a 192/256 ⫻ 512 matrix, a 30- to 45-cm field of view, four excitations, and a half-Fourier postprocessing, was applied. Axial, sagittal, and/or coronal (31 of 42 patients) 5- to 8-mm-thick slice images were obtained, with cardiac gating, interslice gap of 1 to 2 mm. T2-weighted images were obtained in 37 of 42 patients with a nongated turbo spin echo sequence (TR, 7,000 to 7500 ms; expiratory time, 91 ms; eight echotrains; 20 to 25 slices; two excitations; 160/192 ⫻ 256 matrix). In addition, for all patients, contrast-enhanced T1-weighted imaging was obtained after IV administration of 0.2 mL/kg of body weight of Gd-diethylenetriamine pentaacetic acid (Magnevist; Schering AG). Because of technical failure, five patients could not be examined with T2-weighted sequence. The images were assessed for presence of pleural lesions, type, location, and extent of pleural thickening as suggested by previous studies.2,3 The location of the pleural lesions was defined as parietal, visceral, fissural, mediastinal, diaphragmatic, and multiple. Mediastinal or diaphragmatic involvement was defined as pleural thickening bordering the mediastinum or the diaphragm. The distinction between visceral and parietal pleural thickening was made only in the presence of pleural effusion. Diffuse pleural lesions were further characterized by whether it was circumferential (defined as involvement of entire circumference of pleura in any thoracic level) or entire hemithorax (defined as involvement of entire hemithoracic pleura). Pleural thickening was also classified as a pleural thickness of 3 to 10 mm or ⬎ 10 mm. The contour of the pleural thickening was defined as smooth, irregular, and nodular. In CT, infiltration of chest wall was defined as soft tissue infiltration by the tumor and/or lytic destruction of the ribs or vertebrae. MRI criteria of chest wall infiltration was defined as follows: signal intensity identical to that of the tumor

on T1-weighted images, intraparietal hyperintense signal of the tumor on T2-weighted images, and intraparietal enhancement on T1-weighted images after contrast material administration.4 Mediastinal and hilar nodes were considered enlarged if they were ⬎ 10 mm in short-axis diameter. Furthermore, the signal intensity of pleural lesions on each pulse sequence were compared with that of the intercostal muscles. The method of study was retrospective. All imaging studies were read by two experienced radiologists in a consensus mode who were blinded to the final histodiagnosis at the time of reading. Statistical comparison between selected patient groups and selected imaging methods was performed by using the ␹2 test.

Results All 42 patients in the study had pleural thickening. Eight patients had bilateral pleural thickening, and 34 had unilateral pleural thickening. CT features most suggestive of a malignant cause were mediastinal pleural involvement, circumferential pleural thickening, nodularity, irregularity of pleural contour, and infiltration of the chest wall and/or diaphragm (Fig 1, top, A; Table 2). All of these features were significantly more common in malignant than in benign pleural disease (p ⬍ 0.05, ␹2 test). Twenty-five of 27 cases of malignant pleural disease had one or more of these features as compared with only 2 of 15 cases of benign pleural disease, representing a sensitivity of 93% and a specificity of 87% for malignancy. The presence of pleural calcification was suggestive of benign cause, with a sensitivity of 33% and a specificity of 96%. MRI features most suggestive of a malignant cause were mediastinal pleural involvement, circumferential pleural thickening, nodularity, irregularity of pleural contour, and infiltration of the chest wall and/or diaphragm (Fig 1, 2; Table 2). All were significantly more common in malignant than in benign pleural disease (p ⬍ 0.05, ␹2 test). Twenty-six of 27 cases of malignant pleural disease had one or more of these features as compared with only 3 of 15 cases of benign pleural disease, representing a sensitivity of 96% and a specificity of 80% for malignancy. Furthermore, MRI demonstrated additional four more cases of positive chest wall and/or diaphragm infiltration than CT (Fig 2; Table 2). Using the morphologic features, however, MRI and CT had the same diagnostic accuracy of 90% in differential diagnosis of pleural disease in this series. On the basis of comparative evaluation of the magnetic resonance signal intensity of pleural lesions, 20 of 22 cases of malignant lesions appeared to be hyperintense in relation to intercostal muscles on T2-weighted images, as compared with only 3 of 15 cases of benign lesions, representing a sensitivity of 91% and a specificity of 80% for malignancy (86% diagnostic accuracy, 87% positive predictive value, CHEST / 118 / 3 / SEPTEMBER, 2000

605

specificity of 93%, a positive predictive value of 96%, a negative predictive value of 100%, and a diagnostic accuracy of 98% in the detection of pleural malignancy.

Discussion

Figure 1. A 71-year-old patient with pleural mesothelioma. CT scan (top, A) shows circumferential, nodular pleural thickening with mediastinal involvement. The coronal enhanced T1weighted MRI (bottom, B) shows entire hemithorax pleural thickening with nodular, irregular margins. Extensive diaphragmatic and mediastinal pleural involvement can also be seen.

86% negative predictive value). On contrast-enhanced T1-weighted images, 25 of 27 cases of malignant pleural lesions appeared to be hyperintense in relation to intercostal muscles as compared with 4 of 15 cases of benign lesions, representing a sensitivity of 93% and a specificity of 73% for malignancy (86% diagnostic accuracy, 86% positive predictive value, 85% negative predictive value; Fig 3). However, there was no significant difference of signal intensity between malignant and benign pleural lesions on unenhanced T1-weighted images. Using magnetic resonance morphologic features in combination with the signal intensity features, all 27 patients with malignant pleural diseases were correctly identified, representing a sensitivity of 100%, a 606

Pleural space disease is often complex, difficult to diagnose, and problematic to manage. Although chest radiography is always used in the initial assessment of patients with symptoms or signs of pleural pathologic conditions, radiographic findings frequently do not help differentiate malignant from benign pleural disease and parenchymal from pleural processes.3,5 In our study, the CT features most suggestive of a malignant cause were mediastinal pleural involvement, circumferential pleural thickening, nodularity, irregularity of pleural contour, and infiltration of the chest wall and/or diaphragm. An overall sensitivity of 93% and an overall specificity of 87% for malignancy were achieved. All of these features may be seen in mesothelioma, metastatic and other malignant pleural disease but are unusual in benign pleural disease. The presence of pleural calcification was suggestive of a benign cause. However, pleural metastases may appear identical to malignant mesothelioma, characterized by diffuse pleural thickening and encasement of the underlying lung. In such cases, accurate differentiation by CT is difficult.6,7 On the basis of MRI morphologic patterns such as nodularity, pleural thickness ⬎10 mm, mediastinal or circumferential pleural involvement, or pleural extension through the entire hemithorax, malignant pleural disease is suggested; infiltration of the diaphragm and the chest wall were most indicative for malignancy according to the current literature.8,9 In our study, the MRI features most suggestive of malignant disease were mediastinal pleural involvement, circumferential thickening, nodularity, irregularity of pleural contour, and infiltration of the chest wall and/or diaphragm. They had a total sensitivity of 96% and a total specificity of 80% for malignancy. Malignant pleural disease tends to involve the entire pleural surface, whereas reactive pleurisy usually does not affect the mediastinal pleura. The main exception to the rule is tuberculous empyema, which, when extensive, may involve the mediastinal pleura.3,8 In our study, 21 of 27 cases of malignant pleural disease had mediastinal pleural involvement demonstrated on MRI. The sole patient with benign pleural disease showing a mediastinal pleural involvement on MRI was identified pathologically as tuberculous pleurisy. Clinical Investigations

Table 2—CT and MRI Characteristics in 42 Patients With Pleural Diseases Malignant, No. (n ⫽ 27)

Benign, No. (n ⫽ 15)

␹2 Test, p Value*

Imaging Characteristics

CT

MRI

CT

MRI

CT

MRI

Mediastinal involvement Circumferential thickening Nodularity Irregularity Infiltration Calcification

20 8 13 10 8 1

21 8 13 10 12 —

1 0 2 1 0 5

1 0 2 1 0 —

⬍ 0.005 ⬍ 0.025 ⬍ 0.025 ⬍ 0.05 ⬍ 0.025 ⬍ 0.05

⬍ 0.005 ⬍ 0.025 ⬍ 0.025 ⬍ 0.05 ⬍ 0.005 —

*Statistics compare benign vs malignant disease.

MRI, especially when contrast-enhanced T1weighted imaging is applied, has demonstrated its reliable ability in the evaluation of diaphragm and chest wall invasion in patients with different pulmonary masses, or primary or secondary pleural disease.10 Particularly in tumors of the superior sulcus or the lung base, images in the coronal or sagittal plane can be used to demonstrate the relationship of the tumor within the lung apex to adjacent structures. Invasion of the diaphragm can be visualized by contrast enhancement of the regions involved.8 –13 In our study, MRI demonstrated an additional four more cases of positive chest wall and/or diaphragm infiltration than CT. The morphologic features on MRI allowed a largely equal and in selected cases superior detection and evaluation of the spread of pleural disease as compared to CT. However, CT was very sensitive for detecting calcification that was most suggestive of a benign cause. Pleural thickness ⬎ 1 cm was considered as a CT or MRI criterion for malignant pleural disease in the literature.2,3,8,14 In our study, however, this feature had no significant difference between malignant and benign pleural disease (p ⬎ 0.05, ␹2 test). This may be partially due to the more pathologic variety in our study (especially in the benign group), which probably declined the possible false-negative bias in circumstance; and, also, the relatively advanced CT and MRI modalities used in our study, which certainly improved the accuracy of the pleural thickness measurements. In our study, high signal intensity in relation to intercostal muscles on T2-weighted images represented a sensitivity of 91% and a specificity of 80% for malignancy, and represented a sensitivity of 93% and a specificity of 73% for malignancy on contrastenhanced T1-weighted images. However, there was no significant difference of signal intensity between malignant and benign pleural lesions on unenhanced T1-weighted images (p ⬎ 0.05, ␹2 test). The specificities of high signal intensity for malignant pleural disease on T2-weighted and contrast-enhanced T1-

Figure 2. A 61-year-old patient with malignant pleural mesothelioma. The coronal T1-weighted image (top, A) shows pleural mass (arrow) in the right upper thorax. Intraparietal tumoral signal intensity is evident. The right major fissure involvement is also demonstrated. But diaphragmatic infiltration is indistinct. The coronal enhanced T1-weighted image (bottom, B) shows the contrast enhancement of the pleural lesions (arrow). Both the chest wall and diaphragmatic infiltrations (double arrows) are distinguishable.

CHEST / 118 / 3 / SEPTEMBER, 2000

607

weighted images were lower than those observed by Boraschi et al12 and Falaschi et al.14 This may be also partially due to the more pathologic variety, especially the high frequency of tuberculous pleurisy (5 of 42 cases), in our study (compared with only 2 of 34 cases in the study by Falaschi et al14 and none of 30 cases in the study by Boraschi et al,12 who showed a signal intensity in the range of that for malignant pleural lesions).14,15 We must therefore state that magnetic resonance signal intensity is unsuitable as a sole criterion for the differentiation of pleural disease. Using morphologic features in combination with the signal intensity features, MRI had a sensitivity of 100% and a specificity of 93% in the detection of pleural malignancy in our study. The major advantage of MRI over CT lies in its superiority in delineating chest wall and diaphragm invasion. The advantages of CT remain in its sensitive detection of pleural calcification, which is most suggestive of a benign cause, of bone destruction in malignant lesions, and its superior guidance for pleural biopsy.16 Therefore, we recommend that MRI should become the preferable imaging method to assess the extent and resectability of pleural tumor. A major drawback, however, is represented by the cost of MRI which, in our country, is approximately twice as costly as CT. Under these circumstances and under the increasing economic pressure, the value of CT-guided biopsy has to be taken into consideration. A prospective trial has been started in our institute to compare overall diagnostic efficiency, safety, and cost of CT-guided biopsy of pleural disease as compared to imaging protocols alone. This should also illuminate the therapeutic impact of the imaging results, which has not been assessed in the present study, due to the retrospective nature of the study design.

References

Figure 3. MRI of a 58-year-old patient with malignant diffuse pleural mesothelioma. The sagittal T1-weighted image (top, A) shows right extensive pleural thickening with fissural pleural involvement (*). The tumor shows intermediate signal intensity (appears to be isointense in relation to intercostal muscles). In the T2-weighted (middle, B) and enhanced T1-weighted (bottom, C) images, the tumor appears to be hyperintense in relation to intercostal muscles (arrow). 608

1 Roggli VL, Kolbeck J, Sanfillippo F, et al. Pathology of human mesothelioma: etiologic and diagnostic considerations. Pathol Annu 1987; 2:91–131 2 Bittner RC, Scho¨rner W, Loddenkemper C, et al. Pleuraerkrankungen in der Magnetresonanztomographie (MRT). Pneumoligie 1992; 46:612– 620 3 Leung AN, Muller NL, Muller RR. CT in differential diagnosis of diffuse pleural disease. AJR Am J Roentgenol 1990; 154:487– 492 4 Padovani B, Mouroux J, Seksik L, et al. Chest wall invasion by bronchogenic carcinoma: evaluation with MR imaging. Radiology 1993; 187:33–38 5 Scott EM, Marshall TJ, Flower CDR, et al. Diffuse pleural thickening: percutaneous CT-guided cutting needle biopsy. Radiology 1995; 194:867– 870 Clinical Investigations

6 Lewis RJ, Caccavale RJ, Sisler GE. Imaged thoracoscopic lung biopsy. Chest 1992; 102:60 – 62 7 McLoud TC, Flower CDR. Imaging the pleura: sonography, CT and MR imaging. AJR Am J Roentgenol 1991; 156:1145– 1153 8 Mu¨ller NL. Imaging of the pleura. Radiology 1993; 186:297– 309 9 Carlsen SE, Bergin CJ, Hoppe RT. MR imaging to detect chest wall and pleural involvement in patients with lymphoma: effect on radiation therapy planning. AJR Am J Roentgenol 1993; 160:1191–1195 10 Bittner RC, Scho¨rner W, Auffermann W, et al. Contrast enhanced magnetic resonance imaging in chest pathologies [abstract]. Am Rev Respir Dis 1992; 145:A196 11 Bittner RC, Felix R. Magnetic resonance (MR) imaging of the

chest: state-of-the-art. Eur Respir J 1998; 11:1392–1404 12 Boraschi P, Neri S, Braccini G, et al. Magnetic resonance appearance of asbestos-related benign and malignant pleural diseases. Scand J Work Environ Health 1999; 25:18 –23 13 Patz EF Jr, Shaffer K, Piwnica-Worms DR, et al. Malignant pleural mesothelioma: value of CT and MR imaging in predicting resectability. AJR Am J Roentgenol 1992; 159:961– 966 14 Falaschi F, Battdla L, Mascalchi M, et al. Usefulness of MR signal intensity in distinguishing benign from malignant pleural disease. AJR Am J Roentgenol 1996; 166:963–968 15 Roth LTC BJ. Searching for tuberculosis in the pleural space. Chest 1999; 116:3– 4 16 Daly B, Templetom PA. Real-time CT fluoroscopy: evolution of an interventional tool. Radiology 1999; 211:309 –315

CHEST / 118 / 3 / SEPTEMBER, 2000

609