Differentiating lung abscess and empyema: radiography and computed tomographyComputed Tomography w. Richard Webb vision March 3, 1983. 1 All authors: Department of Radiology, Univer- sity of California Medical Center, San Francisco General Hospital, 1 001 Potrero Ave. , San Fran- cisco, CA 941 1 0. Address reprint requests to M. P. Federle. AJR 141:163-167, July 1983 0361-803X/83/141 1-01 63 © American Roentgen Ray Society Conventional chest radiographs and computed tomographic (CT) scans of 70 inflam- matory thoracic lesions in 63 patients were reviewed and scored for diagnostic features. Pathologic confirmation of the final diagnosis was available in 42% (5/1 2) of lung abscesses and 31 % (1 8/58) of empyemas. CT alone was sufficient to correctly diagnose 100% (70/70) of cases. Diagnostic information not available from conven- tional chest radiographs was obtained in 47% (33/70) of cases; in an additional 34% of patients, CT more accurately defined the extent of disease. The most reliable CT features for the differential diagnosis of lung abscess and empyema were wall char- acteristics, pleural separation, and lung compression. Conventional radiographic fea- tures such as size, shape, and the angle of the lesion with the chest wall were less helpful, though also best assessed by CT. The differentiation between lung abscess and empyema can be difficult, but has important therapeutic consequences. Prolonged antibiotic therapy and pos- tural drainage are appropriate for a lung abscess; however, early thoracostomy tube drainage is essential therapy for an empyema [1 , 2]. Chest radiographs are usually relied on to help make the correct diagnosis. The major radiographic criteria for this differential diagnosis are based on assessment of the three-dimensional shape of a peripheral thoracic fluid collec- tion [3]. Overlying lung disease or an unfavorable location of the lesion often results in ambiguous findings. Cl has been used to distinguish lung abscess from empyema as noted in prior reports of small numbers of patients [4, 5]. The Cl criteria have been analogous to those of chest radiographs, primarily shape of the lesion and characteristics of the wall between the lesion and surrounding lung. We undertook this study to assess the relative merits of chest radiographs and Cl in the evaluation of suspected lung abscess or empyema, to define the most reliable radiographic criteria for accurate differentiation, to correlate newly ob- served radiographic features with pathologic findings, and to determine the optimal CT scanning technique for the evaluation of inflammatory thoracic le- sions. Materials and Methods The medical records of 80 patients at the San Francisco General Hospital or the University of California, San Francisco suspected of having a lung abscess or empyema were reviewed. Seventeen ultimately had histologic proof of necrotic lung tumors or malignant pleural effusions and were excluded from further analysis. The 63 remaining patients form the basis for this study, and 70 inflammatory lesions were present in this group. All 63 patients had concurrent plain chest radiographs and chest CT scans. Their medical records, surgical results, on autopsy findings were reviewed to determine the nature of their lesions, the causative organisms, and to determine the grounds for the final D ow nl oa de d fr om w w w .a jr on lin e. or g by 1 71 .2 43 .6 7. 96 o n 01 /1 8/ 23 f ro m I P ad dr es s 17 1. 24 3. 67 .9 6. C op yr ig ht A R R S. F or p er so se o nl y; a ll ri gh ts r es er ve d TABLE 1 : CT Features for the Diagnosis of Lung Abscess vs. Empyema: 70 Lesions in 63 Patients Note-Not all features could be assessed in every patient. . Significant at p < 0.001 , chi-square test, 1 degree of freedom. 164 STARK El AL. AJR:141, July 1983 . .CT Findings Lu ng Abscess % (no.) Empyema % (no.) Wall characteristics: Thick . . . . . . 88 (7/8) 6 (3/54) Thin . . 12 (1/8) 94 (51/54) Width uniformity 0 (0/9) 93 (51 /55) Smooth luminal margin . . 14 (1 /7) 91 (52/57) Smooth exterior . . . . 1 7 (2/i 2) 91 (51 /56) Pleural separation (‘ ‘split pleura”) . 0 (0/10) 68 (39/57) Lung compression 0 (0/1 2) 47 (27/57) Chest wall angle: Acute . . 83 (10/12) 14 (8/57) Obtuse . . . . . . 0 (0/12) 70 (40/57) Both . . . . . . . 17 (2/12) 16 (9/57) Shape: Round . . . . . 67 (8/1 2) 1 1 (6/56) Oblong 33 (4/12) 27 (15/56) Lenticular 0 (0/12) 63 (35/56) Size: Small . 83 (1 0/1 2) 23 (13/56) Medium . . 8 (1 /1 2) 59 (33/56) Large 8 (1/12) 18 (10/56) Air in lesion . . 75 (9/i 2) 56 (32/57) Lung consolidation . . . 92 (1 1 /1 2) 81 (42/52) Free pleural fluid . 25 (3/1 2) 1 7 (9/52) Septated lesions . 25 (3/1 2) 1 8 (10/57) Multiple lesions . . . 25 (3/i 2) 1 6 (1 1 /51) diagnosis of lung abscess on empyema. Chest radiographs were obtained either with a mobile x-ray unit at low kilovoltage technique on with standard postenoantenior and lateral upright 1 40 kV technique. CT examinations were performed using a GE 8800 scanner, 4.8 or 5.7 scan times, after the intrave- nous injection of 1 50 ml of 60% meglumine diatnizoate in most cases. Suspension of respiration during scanning was often not possible. Contiguous 1 cm slices from lung apex to lung base were performed and photographed at two window and level settings, one optimal for soft tissues (level 500 H, width 20-60 H), the other optimal for lung parenchyma (level 1 000 H, width -550 to -650 H). Without knowledge ofclinical data, laboratory tests, or pathologic findings, 63 CT scans were reviewed by three of us (D. S., M. F., P. G.) and 70 lesions were scored by consensus for 14 diagnostic features (table 1 ). Wall characteristics of the lesions were assessed at the boundary between the lesion and the lung parenchyma or the chest wall. When possible, the luminal margin (inner wall) was examined where the lesion contained gas, and the wall exterior was examined where the adjacent lung was not consolidated. Lesions occupying more than 40% of the hemithorax were considered large, 1 O%-40% medium, and less than 1 0% small. All criteria could not be scored in every case due to adjacent lung disease on technical factors. The plain chest nadiognaphs were then reviewed and com- pared to the CT scans in order to assess their relative value in making the diagnosis of abscess on empyema and their ability to define the extent of disease. Statistical analysis, where appropriate, was done using the chi- square test and standard tables [6]. Results There were 58 empyemas and 1 2 abscesses. CT findings alone were sufficient to make the correct diagnosis of lung abscess or empyema in all 70 cases. Multiple diagnostic CT features were present in even the most challenging case. Proof of the final diagnosis was based on direct (autopsy or open thoracotomy) pathologic examination of the lesion in 33% (23/70) of cases (5/ 1 2 of lung abscesses and 18/58 of empyemas). Thoracostomy tube drainage confirmed the diagnosis of empyema in 53% of cases. Strong clinical evidence, including resolution of abscesses on antibiotic therapy, confirmed the remaining cases. CT features of lung abscess or empyema were similar in pathologically proven and clinically diagnosed cases; therefore, data were com- bined for all patients with a final diagnosis of lung abscess and compared to all patients with a final diagnosis of em- pyema (table 1). Separation of uniformly thickened visceral pleura from parietal pleura was seen in 68% of all empyemas and in an even higher percentage when intravenous contrast material had been administered (fig. 1 ). Compression of uninvolved lung was identified by noting distorted and bowed bronchi and/or pulmonary vessels around the periphery of the lesion (figs. 2 and 3C). Large empyemas were more likely to exhibit lung compression (60% [8/1 4]) than medium (55% [18/ 33]) or small (1 0% [1 /1 0]) empyemas. Pleural separation, called the ‘ ‘split pleura’ ‘ sign, and lung compression were specific features for empyema, as they were never seen in a lung abscess (table 1). Wall characteristics also were helpful, as empyemas reli- ably (94% [51 /54]) had at least a part of their wall that was distinctly thin, uniform, and smooth on both its luminal margin and exterior surface (figs. 1 -3). Some wall charac- teristics could not be evaluated accurately in 7% (4/58) of patients with empyema because of extensive associated lung consolidation. Lung abscesses were typically small with thick, irregular walls relative to the size of the lesion (fig. 4). CT photography at ‘ ‘lung windows’ ‘ was essential to distinguish trapped normal lung surrounded by pneumonia from a cavitating lung abscess (fig. 4). In one case, a large empyema resulted from pleural extension of a small lung abscess not detectable with ‘ ‘ soft-tissue windows’ ‘ alone. Lesion size, shape and chest wall angle were less reliable features in our series. Abscesses tended to be round with acute chest wall angles and empyemas lenticular with ob- tuse chest wall angles; however, acute angles were found in i 4% of empyemas and even coexisted with other areas having obtuse chest wall angles in 1 6% of patients. A nondifferentiating finding was associated lung consolida- tion, present in 92% of patients with lung abscess and 8i % of patients with empyema. Many of these features are physiologically interdependent and were closely associated. For example, wall thinness and smoothness confirmed the finding of wall uniformity, which by itself was diagnostic of 93% (51 /55) of empyemas (table 1). D ow nl oa de d fr om w w w .a jr on lin e. or g by 1 71 .2 43 .6 7. 96 o n 01 /1 8/ 23 f ro m I P ad dr es s 17 1. 24 3. 67 .9 6. C op yr ig ht A R R S. F or p er so Fig. 1 -A, Chronic empyema at left costophrenic angle with marked and obtuse angles with chest wall. Proximal rib (arrowhead). B, Pathologic pleural thickening. Separation of visceral pleura (straightarrow) from panietal specimen. Thickened hypervascular visceral pleura (P). Interface with lung pleura (curved arrow): ‘ ‘split pleura’ ‘ sign. Empyema has lenticular shape (arrows). AJR:141, July 1983 LUNG ABSCESS & EMPYEMA ON RADIOGRAPHY AND CT 165 Fig. 2.-Large empyema is round in cross section and forms acute angles with chest wall. Compressed lung and distorted bronchi (arrowhead) are draped over empyema. Luminal margin Is smooth and ‘ ‘split pleura’ ‘ sign is present. Visceral pleura (straight arrow); parietal pleura (curved arrow). The most common indication for CT scanning was the inability of initial clinical data, laboratory tests, and plain chest radiographs to distinguish lung abscess from em- pyema. CT provided significant additional diagnostic infor- mation in 47% (33/70) of cases and, in this retrospective study, always led to the correct diagnosis (1 00% accuracy). Plain chest radiographs were usually adequate to assess the extent of a lung abscess or an empyema; however, in 34% (24/70) of cases, CT more accurately delineated the extent of disease. Adjacent lung compression or consoli- dation was reliably distinguished from the lesion itself (fig. 2). In addition, CT often detected unforeseen complications, not apparent on the plain chest radiograph, such as thora- costomy tube malpositioning, pneumothorax, contralateral lung disease, mediastinal involvement, or osteomyelitis (fig. 3). Finally, CT accurately delineated chest wall pathology resulting from a chronic empyema, from extension of an actinomycotic abscess, or from trauma that subsequently resulted in empyema. agent was possible. A bacteriologic diagnosis was made in 49% (34/70) cases. Gram-positive organisms (Staphylo- coccus, Streptococcus) were responsible for 50% (4/8) of lung abscesses and 38% (1 0/26) of empyemas. Gram- negative organisms were not implicated in our cases of lung abscess but were the second most common cause of em- pyemas (27% [7/26]). Anaerobic bacteria, mixed infections, tubercle bacillus, and actinomycosis accounted for 38% (1 3/34) of lung abscesses and empyemas combined. Discussion peripheral thoracic lesions have been reported as being useful for differentiating lung abscesses and empyemas [3- 5]. Our data confirm these authors’ conclusions that ab- scesses tend to be round and empyemas lenticular; how- ever, exceptions often occur, and even lenticular empyemas may appear rounded on selected axial CT sections (fig. 3). This study demonstrates the importance of other, more specific CT criteria for the differential diagnosis of lung abscess and empyema. Fig. 3.-Empyema. Posteroanterior (A) and lateral (B) chest radiographs. Medium-sized peripheral fluid collection with smooth outer wall and smooth luminal margin outlined by free gas within lesion. C and D, CT scans. Lesion appears round with acute chest wall angles on axial sections. Compressed lung with bowed and distorted bronchi (arrowhead) is seen above gas- A lung abscess is defined as a localized area of suppur- ation with destruction of lung parenchyma [2]. Cross-sec- tional CT images are ideally suited to demonstrate these lesions as round, thick-walled cavities in areas of destroyed lung. Bronchi and pulmonary vessels terminate abruptly at the advancing wall of an abscess and are not compressed or distorted. Abscess walls are typically irregular in width and have rregular luminal margins and exterior surfaces (figs. 4 and 5). Infected bullae may mimic an abscess, containing lesion with smooth luminal margin. Visceral pleura (arrows). Multiple additional CT findings are not apparent on chest radiographs. Gas within mediastinal abscess (open arrow) and adjacent thoracic vertebra, indicating osteomyelitis. Two additional empyemas in left hemithorax, one immediately posterior to aorta and other along posterolateral thoracic wall. especially when there is adjacent pneumonia; however, a smooth luminal margin on plain film [7] or CT suggests the correct diagnosis. Thoracic empyema is defined as pus in the pleural cavity [2]. As an empyema progresses, a fibrin peel coats the visceral and parietal pleural surfaces. This peel organizes with the ingrowth of capillaries and fibroblasts as early as 7 days after the onset of disease [2], forming the basis for the most reliable radiographic feature of an empyema, which D ow nl oa de d fr om w w w .a jr on lin e. or g by 1 71 .2 43 .6 7. 96 o n 01 /1 8/ 23 f ro m I P ad dr es s 17 1. 24 3. 67 .9 6. C op yr ig ht A R R S. F or p er so se o nl y; a ll ri gh ts r es er ve d AJR:141, July 1983 LUNG ABSCESS & EMPYEMA ON RADIOGRAPHY AND CT 167 acteristic of an abscess. Empyemas also may be distin- guished when they have exclusively obtuse angles with the chest wall, lenticular shape, and uniform wall characteris- tics, and when they compress the adjacent lung. Supine and decubitus CT scanning has been recom- mended to evaluate gravitational changes in shape of a peripheral lesion, suggesting an empyema [4]. In our expe- rience, this sign is present only in large empyemas which have many other diagnostic features. Our results suggest that a single series of supine CT sections, using 1 cm slices and intravenous contrast material, is sufficient to character- ize and delineate the full extent of most lung abscesses and empyemas. When clinical findings and conventional radiographic ex- aminations are ambiguous, Cl can accurately diagnose and influence the management of patients with suspected lung abscess or empyema. Fig. 4.-Lung abscess. CT scan at “lung window” setting (-600 H level, 1 000 H width). Small cavitary lesion forms acute angles with chest wall (arrow). Luminal margin is characteristically thick and irregular. No lung compression is present, and undistorted pulmonary vessels end abruptly at lesion wall. Using ‘ ‘soft-tissue windows’ ‘ alone, abscess cavity could not be distinguished from uninvolved lung surrounded by pneumonia. we call the ‘ ‘split pleura’ ‘ sign (fig. 1 ). Visualization of thickened, separated visceral and parietal pleural surfaces C ‘split pleura’ ‘) was noted in 68% of all empyemas. This sign was present in a higher percentage of cases in which intravenous contrast material had been administered, due to contrast enhancement of the pathologically hypervascular pleural surfaces. Rarely the wall of an abscess may abut the pleural surface and mimic pleural thickening; however, the luminal margin will always have the irregularities char- REFERENCES 1 . Bartlett JG, Finegold SM. Anaerobic infections of the lung and pleural space. Am Rev Respir Dis 1 974; 1 1 0 : 56-77 2. Alexander JC, Wolfe WG. Lung abscess and empyema of the thorax. Surg Clin North Am 1980;60:835-849 3. Friedman PJ, Hellekant CAG. Aadiologic recognition of bron- chopleural fistula. Radiology 1 977;1 24 : 289-295 4. Baben CE, Hedlung LW, Oddson TA, Putman CE. Differentiat- ing empyemas and peripheral pulmonary abscesses. Radiology 1980;1 35: 755-758 5. Pugatch AD, Faling U, Robbins AH, Snider GL. Differentiation of pleural and pulmonary lesions using computed tomography. J Comput Assist Tomogr 1 978;2 :601-606 6. Zan JH. Biostatistical analysis. Englewood Cliffs, NJ: Prentice Hall, 1974:60 7. Stark P, Gadziala N, Greene A. Fluid accumulation in pre- existing pulmonary air spaces. AJR 1 980; 134:701-706 D ow nl oa de d fr om w w w .a jr on lin e. or g by 1 71 .2 43 .6 7. 96 o n 01 /1 8/ 23 f ro m I P ad dr es s 17 1. 24 3. 67 .9 6. C op yr ig ht A R R S. F or p er so
LOAD MORE