Diagnosis of Pericardial Effusion by Computed Tomography

Diagnosis of Pericardial Effusion by Computed Tomography* Bert Y. S. Wong, M.D., F.G.G.P.; Kyo R. Lee, M.D.; and Richard I. MacArthur, M.D.

w.

We assessed the diagnostic utility of computed tomography (Cf) in the detection of pericardial effusion by using a dog model in which the amount and type of pericardial flald were controDed and by obtaining cr scans on patients with suspected pericardial effusion prior to pericardiocentesis. The experimental studies showed that CT was capable of detecting pericardial effusion composed of serous luid or saline in amounts as small as SO mL However, the hemopericardium was

radiographically WocIense with the heart and difticult to detect by CT unless the epicardial fat pad identified or intravenous contrast enhancement of the heart was used. CT clearly demonstrated the presence of pericardial etl'usioD in all six patients studied prior to successful pericanUocentesis or pericardiectomy. We conclude that CT is a sensitive noninvasive method which can be used for the diagnosis of pericardial etl'usion.

Making a definitive diagnosis of pericardial effusion purely on clinical grounds is often difficult. The clinical signs of cardiac tamponade are not pathognomonic of pericardial fluid and may be present in the absence of pericardial effusion. 1,2 Large amounts of pericardial effusion may be present without causing signs of cardiac tamponade," Presently, the best noninvasive method for detecting pericardial effusion is echoeardiography." Echocardiography is limited because both false-positive and false-negative echocardiograms for determining the presence of pericardial effusion can occur and because it is not always possible to obtain an interpretable echocardiogram of every patient.P" Thus, an alternative noninvasive method for diagnosing pericardial effusion would be clinically valuable. We undertook this study to assess the diagnostic capability of computed tomography ( CT) for detecting pericardial effusion,

ously at the dorsal base of the neck and attached to a stopcock. This allowed the instillation and removal of fluid from the pericardal sac. A total of ten experiments were performed on seven dogs. In three experiments done immediately after the placement of the pericardial catheter, CT studies were performed before and after instillation of 50-100 ml of normal saline solution in the pericardial sac. In seven experiments carried out four days to three weeks after the pericardial tube was placed, chronic pericardial effusion was produced by the daily instillation of 50 ml of 50 percent dextran. CT scans were obtained before and after removal of a known amount of pericardial fluid. Additionally, in three of these seven experiments, CT scans were obtained after the chronic pericardial effusion was replaced by heparinized whole blood. CT scans of the hemopericardium were taken both with and without intravenous contrast enhancement (150 ml of 30 percent methylglucamine diatrizoate given by the drip infusion technique). A total of 30 CT studies were performed: ten control studies of hearts with no pericardial fluid, seven of hearts with 50-100 ml of pericardial fluid, five of hearts with 100-300 ml of pericardial fluid, two of hearts with more than 300 ml of pericardial fluid, and six of hearts with 200-300 ml of hemopericardium-three without intravenous contrast enhancement and three with intravenous contrast enhancement. All dogs were studied while under general anesthesia with intravenous pentobarbital with an endotracheal tube in place, but with spontaneous respiration. We confirmed that aspiration could remove virtually all the pericardial fluid (within 15-20 ml) at the time of the initial surgery in all dogs and at postmortem examination in two dogs with chronic pericardial effusion. Additionally, at the end of each experiment, CT scanning was repeated after instilling 50 ml of diluted contrast solution into the pericardial sac to be sure that the experimental system was intact and that there was no extravasation of fluid out of the pericardial sac. Second, CT studies of the chest were obtained in six patients within one week prior to pericardiocentesis or pericardiectomy. These patients were suspected to have a pericardial effusion on clinical grounds. Five of the six

METHOD

This study was carried out in two ways. First.. we used an experimental dog model in which the amount and type of pericardial fluid were controlled. Under general anesthesia with intravenous pentobarbital (25 mg/kg ), one end of a large bore (0.104 ID) flexible plastic tube was inserted into the pericardial sac through a left lateral thoracotomy and sutured in place. The other end of the tube was brought out from the thoracotomy incision subcutane°From the Departments of Medicine, Diagnostic Radiology, and Surgery, University of Kansas Medical Center, Kansas City, Kansas. Presented in part at the 45th Annual Scientific Assembly, American College of Chest Physicians, Houston, November 4-8,1979. Reprint requests: Dr. Wong, University of Kansas Medical Center, 39th and Rainbow, Kansas City, KS 66103

CHEST, 81: 2, FEBRUARY, 1982

DIAGNOSIS OF PERICARDIAL EFFUSION 177

patients had echocardiograms indicative of pericardial effusion. The patient with an echocardiogram negative for pericardial effusion was suspected to have a subacute effusive constrictive pericarditis on the basis of an increase in the size of the cardiac silhouette by chest roentgenogram and the development of a positive Kussmaul sign on physical examination. This patient at surgery turned out to have tumor encasement with a small amount of pericardial effusion. CT scans were taken with a GE CT IT 8800 body scanner with 4.8 sec scanning time and 1 em slice thickness . Scans were obtained at 1 em intervals from the suprasternal notch to the xyphoid process . EXPERIMENTAL REsULTS

All pericardia! effusions composed of saline solution or serosanguinous fluid of 50 ml or more were easily detected by CT. Small amounts of fluid in the range of 50-100 ml usually did not surround the heart, but would layer in the dependent areas (Fig 1) . Multiple scans of the heart were required to detect the presence of this small amount of fluid, as individual scans could miss the area in which the fluid was layered. With 200 ml of fluid, the heart was almost completely surrounded by the fluid and 300 ml of pericardial fluid completely surrounded the heart. Pericardial effusions of 200 ml or more

were easily detected on almost all individual scans which included the heart. The CT studies of hemopericardium showed that blood was isodense and difficult to separate from the heart (Fig 2A, C). However, identification of the epicardial fat pad which showed a clear ring around the heart could be used to separate the heart from the hemopericardium (Fig 2B). Additionally, contrast enhancement of the heart with an intravenous infusion of contrast material also helped in separating the heart from the hemopericardium (Fig 2D). RESULTS OF CLINICAL STUDIES

The results of CT scans obtained on six patients prior to pericardiocentesisipericardiectomy are shown in Table 1. Pericardial fluid was detected in all patients by CT. The amount of pericardial fluid removed ranged from 150 to over 900 ml. All the patients had either serous or serosanguinous fluid. The protein in the fluid ranged from 2.4 to 6.1 mg percent. The transverse CT scans of two patients, one with pericardial effusion only and the other with both pericardial and pleural effusions, are shown in Figure 3, and the reconstructed sagit-

1. CT scans of a dog with pericardial Huid. A. Scan after placement of a catheter in the pericardial sac without injection of fluid shows a very thin , lucent layer (arrow) around part of the heart (H) . B. Scan with 50 ml of saline solution shows an easily visible lucent layer (arrow) around the lower portion of the heart (H). The CT number of the heart is 20. C. Scan with 100 ml of saline solution shows a thick, lucent layer (arrow) encircling the heart (H) . At, BI, Ct . The lucent layers in A, B, and C show a CT number of 0 (fluid density) and are enhanced photographicaIly with Hashing technique (arrows). FIGURE

178 WONG ET AL

CHEST, 81: 2, FEBRUARY, 1982

FIGURE 2. CT scans of three different dogs with hemopericardium (A, B, C). D is a postcontrast scan of C. The hemopericardium and heart cannot be distingu ished (A and C) unless the pericardial fat pad (arrows in B) is identified or the heart chambers are enhanced by the contrast material (asterisks ill D).

tal view of the second patient is shown in Figure 4. The CT scan of patient 6 was quite unusual. This patient had a small amount of loculated pericardial fluid with an unusual configuration of the heart by CT scan. This patient had carcinoma of the lung and at pericardiectomy was found to have tumor invasion of the pericardium. The tumor on CT scan was isodense with the heart and could not be separated from the heart. However, the unusual configuration of the cardiac image suggested that a tumor was present. Neither the M mode nor the sector echocardiograms of this patient suggested pericardia] effusion or tumor. Table l--Clinical Findinga and Computed Tomography; Reaul,. 0/ Patien" with Suapected Pericardial Effuaiona

Patient No . Age/Sex

Fluid Removed TP Amount Diagnosis (rnl) Het (mll: %)

34, :\f

Viral

BOO

3

4.7

2

60, F

Uremia

BOO

2

3.7

3

58, F

Viral

200

4

30, F

The

200

5

60, :\f

Tumor"

150

6

82, F

Malignant "

350

2.4 4

6.1

5.3

"Confirmed [Positive for perieardial fluid TP = total protein, Hct = hematocrit , CT = eornpu ted tomography

CHEST, 81: 2, FEBRUARY, 1982

CT

+ + + + + +

FIGURE 3. CT scans of two patients (A) with pericardial effusion alone and (B) with pericardial effusion and bilateral pleural effusions . Transverse scan demonstrates the pericardium (arrows) clearly separating the pericardial effusion (e) from co· existing pleural effusion (e') . H Heart, L collapsed left lower lung, v inferior vena cava, a aorta, S spine.

= =

=

=

=

DIAGNOSIS OF PERICARDIAL EFFUSION 179

4. Reconstructed sagittal scan at the left para-spinous line demonstrates the heart ( H) surrounded by pericardial effusion (e) . The arrows indicate the epicardial fat pad. The pleural effusion is seen in the posterior and lower chest (e") . L = collapsed left lower lung, L' left lobe of the liver , S spine. FIGURE

=

=

DISCUSSION

There have been anecdotal reports in which pericardial effusion was diagnosed by CT.8.IO However, the sensitivity and specificity of CT for the diagnosis of pericardial effusion has not been defined previously. One author has stated that even large pericardial effusions would be difficult to diagnose because of movement artifact. II Our study indicates that CT is a very sensitive method which can detect pericardial effusion in amounts as small as 50 ml. Movement of the thorax with normal respiration did not present any significant motion artifact either in the animal experiments or in patients. Cardiac movement affected the CT of the heart itself and made it difficult to clearly identify individual chambers, but it did not affect the identification of the pericardial fluid which surrounded the heart. The degree of sensitivity of CT for diagnosing pericardial effusion (detecting 50 ml or more) found in this study is quite comparable to what has been reported for echocardiography.'>" The specificity of CT for diagnosis of pericardial effusion depends on the type of effusion present. Our studies indicate that both transudative and exudative fluids can be distinguished easily from the heart with CT by the difference in CT numbers

180 WONG ET AL

(Fig 2). However, the experimental studies clearly showed that hemopericardium was isodense with the heart and thus difficult to separate from the heart by CT unless the epicardial fat pad'" was identified or intravenous contrast enhancement of the heart was used. Thus , it is likely that hemopericardium in patients will also be difficult to detect by CT . Our experience with one patient with tumor in the pericardium shows that intrapericardial tumor can also be difficult to diagnose by CT, as it was isodense to the heart on the CT scan. This study has shown that CT is a sensitive method for detecting pericardial effusion in an experimental model and that it can be used to detect pericardial effusion in patients. Thus, clinically it should prove to be a valuable adjunct for the noninvasive diagnosis of pericardial effusion. However, further studies are required to determine the sensitivity and specificity of the method in a large number of patients. At this time, we do not recommend that this method be used in place of echocardiography. It has its own limitations in that it cannot be used at the bedside and it also exposes the patient to a certain amount of radiation (approximately 1-2 rads) .17 However, we do recommend that it be used as an alternative noninvasive

CHEST, 81: 2, FEBRUARY, 1982

method for diagnosing pericardial effusion if echocardiography does not provide a definitive diagno-

sis.

ACKNOWLEDGMENT: We appreciate the technical assistance of Ronald Smith, and the secretarial assistance of Carol Scott. REFERENCES

1 Spodiek DH. Acute cardiac tamponade. Pathologic physiology, diagnosis and management. Prog Cardiovasc Dis 1967; 20: 64-96 2 Fowler NO. Physiology of cardiac tamponade and pulsus paradoxicus. Mod Concepts Cardiovasc Dis 1978; 47: 109-13 3 Riba AC, Morganroth ]. Unsuspected substantial pericardia! effusion detected by eehocardiography. ]AMA 1976; 236:2623-25 4 Feigenbaum H. Echocardiographic diagnosis of pericardia! effusion. Am J Cardiol 1970; 26:475-79 5 Jacobs WR, Talano JV, Loeb HS. Echocardiographic interpretation of pericardial effusion. Arch Intern Med 1978; 138:622-2.5 6 Millman A, Meller ], Motro M, Blank HS, Horowitz I, Herman MV, Teicholz LE. Pericardia! tumor or fibrosis mimicking pericardial effusion by echocardiography. Ann Intern Med 1977; 86:434-36 7 Foote WC, Jefferson CM, Price HL. False positive echocardiographic diagnosis of pericardial effusion. Chest 1977; 71:546-49 8 Alfidi RJ, Haaga I, Meaney TF, MacIntyre WI, Gonzalez

CHEST, 81: 2, FEBRUARY, 1982

L, Tarar R, Zelch MG, Boller M, Cook SA, ]elden G. Computed tomography of the thorax and abdomen; a preliminary report. Radiology 1975; 117:257-64 9 Stanley R], Sagel SS, Levitt RG. Computed tomography of the body: early trends in application and accuracy of the method. Am I Roentgenol Radium Ther Nucl Med 1976; 127:53-67 10 Abrams HL, McNeil BJ. Medical implications of computed tomography ("cat scanning"). N Engl I Med 1978; 298:255-61, 310-17 11 Kreel L. Computed tomography and the cardiovascular system. In: Dickinson CJ, Marks J, eds. Developments in cardiovascular medicine. Baltimore: University Park Press 1978:3-9 12 Feigenbaum H, Waldhausen ]A, Hyde LP. Ultrasonic diagnosis of pericardia! effusion. JAMA 1965; 191: 107 13 Christiansen E, Bonte F. The relative accuracy of echocardiography, intravenous CO 2 studies, and blood pool scanning in detecting pericardial effusions in dogs. Radiology 1968; 91:265 14 Klein II, Ruber G, Shimada H, Kingsley B, Segal B. Evaluation of induced pericardia! effusion by reflected ultrasound. Am I Cardiol 1968; 22:49 15 Horowitz MS, Schultz CS, Stinson EB, Harrison DC, Popp RL. Sensitivity and specificity of echocardiographic diagnosis of pericardial effusion. Circulation 1974; 50: 239-47 16 Carsky ZN, Mauceri RA, Azimi F. The epicardial fat pad sign. Radiology 1980; 137:303-08 17 McCullough EC, Payne JT. Patient dosage in computed tomography. Radiology 1978; 129:457-63

DIAGNOSIS OF PERICARDIAl EFFUSION 181