Delayed Postoperative Cardiac Tamponade: Diagnosis and Management

Transcription

1 Delayed Postoperative Cardiac Tamponade: Diagnosis and Management Robert L. Hardesty, M.D., Mark Thompson, M.D., David B. Lerberg, M.D., Ralph D. Siewers,-M.D., James D. O Toole, M.D., Rosemarie Salerni, M.D., and Henry T. Bahnson, M.D. ABSTRACT Symptoms and signs of decreased cardiac output associated with an elevated venous pressure should alert one to the possibility of delayed cardiac tamponade. Enlargement of the cardiothoracic ratio shown by serial roentgenograms and demonstratipn of significant pericardial effusion by echocardiogram or radionuclide angiocardiography support the diagnosis. Erratic response of the prothrombin time to administration of warfarin and abnormal results of liver function test are additional clues to its diagnosis. Right heart catheterization documents the presence of tamponade and excludes other diagnbstic considerations. Operative decompression of the pericardial space can be accomplished by pericardicentesis, subxiphoid pericardiotomy, hedian sternotomy, or thoracotomy. Hemodynamic observations following the relief of tamponade assure that an adequate therapeutic procedure has been performed. Delayed cardiac tamponade that occurs late following a cardiac operation is not anticipated, readily apparent, or easily recognized. Its clinical presentation develops insidiously during the postoperative period and is manifested by low cardiac output. When it is diagnosed, decompression of the pericardial space may be achieved by pericardicentesis, subxiphoid pericardiotomy, repeat median sternotomy, or thoracotomy. Experience with this complication has elucidated some principles regarding diagnosis and management that warrant emphasis. Clinical Material At the Hospitals of the University Health Center of Pittsburgh, 1,249 cardiac operations From the Departments of Surgery and Cardiology, University of Pittsburgh School of Medicine, Pittsburgh, PA. Accepted for publication Nov 22, Address reprint requests to Dr. Hardesty, Department of Surgery, University of Pittsburgh School of Medicine, Pittsburgh, PA employing cardiopulmonary bypass were performed in the four-year interval from January, 1973, to December, During recovery, 8 patients were confirmed to have had cardiac tamponade within 6 to 21 days following operation; 2 patients who were evaluated for tamponade were found to have hemopericardium and no tamponade. A large number of the operations employing extracorporeal circulation were performed on infants and children being treated for congenital heart disease, yet it was not suspected that any of these patients had delayed tamponade. The patients with delayed tamponade ranged in age from 26 to 55 years and underwent an operation for acquired heart disease, with prosthetic valve replacement in 8, mitral commissurotomy in 1, and coronary artery bypass grafting in 1. A loose closure of the pericardium is commonly done but was not employed during these 10 operations. In each instance the mediastinum was drained by gravity with a large (32F) tube placed anterior to the heart and a second placed posteriorly within the pericardial space, with the tip adjacent to the left atrial appendage. Patients who had a prosthetic valve inserted were started on a warfarin (Coumadin) regimen beginning 48 hours after operation. The patient who underwent mitral commissurotomy was also started on warfarin beginning 48 hours after the procedure. The patient who had coronary artery bypass was not given anticoagulants. The clinical presentation and laboratory data were documented by a review of the hospital records. The diagnosis of delayed cardiac tamponade was confirmed by cardiac catheterization in 5 patients. Cardiac tamponade was diagnosed when the pulmonary artery wedge, pulmonary artery end-diastolic, right ventricular end-diastolic, and mean right atrial pressures were equal. In the remainder of the pa by Robert L. Hardesty

2 156 The Annals of Thoracic Surgery Vol 26 No 2 August 1978 tients the diagnosis was confirmed by operative decompression of the pericardium that was followed by hemodynamic improvement, including an increase in the mean systemic arterial pressure, a decrease in the mean central venous pressure, and a clinical judgment that cardiac output had improved. Symptoms heralding decompensation due to delayed tamponade preceded its recognition by two to four days and were initially attributed to a slow recovery and poor ventricular function. These consisted of weakness in all 8 patients, malaise (8/8), dyspnea (8/8), chest pain (7/8), and anorexia (718). Signs suggesting low cardiac output were diaphoresis (8/8), elevation of the central venous pressure (8/8), tachycardia (8/8), tachypnea (8/8), and hepatomegaly (616). A paradoxical pulse (618) and hypotension (818) occurred late and immediately preceded an accurate assessment of the cause for the preceding nonspecific symptoms and signs. Definite trends in laboratory data have been identified that facilitate the recognition of impending tamponade before pulsus paradoxus and hypotension ensue. Serial observations of prothrombin time and roentgenographic assessment of the cardiac silhouette provide early evidence of impending tamponade. Excessive prothrombinopenia relative to the amount of warfarin administration (618) was an early clue: the prothrombin times were both extremely variable and unpredictable. This was associated with abnormally high serum concentration of serum glutamic oxaloacetic transaminase (SGOT), gamma glutamic pyruvic transaminase (ygpt), serum glutamic pyruvic transaminase (SGPT), lactic dehydrogenase (LDH), and bilirubin when hepatic function tests were obtained (315). An unexplained decrease in hemoglobin and hematocrit occurred several days or hours before the recognition of tamponade (7/8). Serial roentgenograms exhibited an unusual and unexpected degree of cardiomegaly relative to the cardiothoracic ratio prior to operation (618). A rapid increase in the cardiothoracic ratio frequently corresponded with an excessive prothrombinopenia and decline in serum hemoglobin and hematocrit (6/8). The voltage of the electrocardiograms was not inordinately low (8/8). No patient received a drug known to be synergistic with warfarin. Data extracted from our experience with 2 patients is especially illustrative of the trends in laboratory data that have been identified. EXAMPLE 1. A 47-year-old patient underwent insertion of a Bjork-Shiley prosthesis (Fig 1A). Excessive bleeding ensued, but immediate tamponade did not occur, the bleeding stopped, and reoperation was not thought to be necessary. Figure 1B shows a significant increase in the cardiothoracic ratio during the first 24 hours following operation. These roentgenograms suggest the presence of marked hemopericardium even though recovery was satisfactory. Figure 1C suggests progression of the patient s cardiac dilatation or effusion on the thirteenth postoperative day, when the clinical course indicated that his cardiac output was lower than expected. During the interim from the thirteenth to the sixteenth day following operation, an unexpected and variable response to the administration of warfarin, an excessive prothrombinopenia, a decline in the serum hemoglobin concentration and hematocrit, and a concomitant increase in the cardiothoracic ratio (Fig 1D) occurred. An echocardiogram confirmed a suspected large pericardial effusion. A paradoxical pulse, mild hypotension, and poor peripheral perfusion were evident. The diagnosis of delayed tamponade was confirmed by the presence of a very large, bloody pericardial effusion that was drained by a subxiphoid pericardiotomy, followed by improvement in his hemodynamic status. EXAMPLE 2. Bjork-Shiley prostheses were inserted in the aortic, mitral, and tricuspid positions in a 38-year-old woman. An inordinately low volume (250 ml) of blood drained from the mediastinal tubes over the 48 hours subsequent to operation. Her immediate recovery was satisfactory. One week after the operation her liver was palpable 6 cm below the right costal margin, as it had been prior to operation, and her hepatic function test results were as follows: total bilirubin, 2.9 mglloo ml; direct bilirubin 1.4 mg1loo ml; ygpt 1.2 X normal; SGPT 1.1 x normal; SGOT 1.6 X normal; and LDH 1.4 x normal. Beginning 72 hours after operation, small doses of warfarin were administered for 10 days (Fig 2); none was given

3 157 Hardesty et al: Delayed Postoperative Cardiac Tamponade C D Fig 2. Chest roentgenograms (A) following insertion of a Bjork-Shiley aortic prosthesis, (B) 24 hours postoperatively, (C) 13 days postoperatively, and (D) 26 days postoperatively.

4 158 The Annals of Thoracic Surgery Vol 26 No 2 August G W c" E tn P 8 I + 0 a n 60 - x % I 2 POST-OP DAYS 4 12 Fig 2. Erratic response of prothrombin time to the administration of warfarin (Coumadin) following triple valve replacement. subsequent to this, yet the prothrombin time continued to rise to a hazardous value, increasing from 60 seconds to 80 seconds over a threehour interval on the twelfth day after operation. During that three-hour period her hematocrit decreased from a previously stable value of 44% to 38%. Her cardiothoracic ratio had increased significantly from the tenth to the twelfth day after operation; previously, it had always been as large as would have been anticipated following a triple valve replacement. Acute hypotension and bradycardia ensued, and a large hemopericardium was evacuated through a subxiphoid pericardiotomy, with the procedure carried out in her hospital room. Her eventual recovery was slow but successful. When clinical and laboratory data make delayed cardiac tamponade a likely consideration, and noninvasive studies suggest the presence of pericardial effusion or intrapericardial thrombus, right heart catheterization has been a valuable procedure to confirm the diagnosis of cardiac tamponade and to aid in the management. This was done in our patients. In each patient the pulmonary artery wedge pressure, pulmonary artery end-diastolic pressure, right ventricular end-diastolic pressure, and mean right atrial pressure were equal. Contours char- acteristic of tamponade were recorded from the right atrial and right ventricular cavities. The right ventricular end-diastolic pressure was equal to or greater than a third of the peak systolic right ventricular pressure. The y descent in the atrial pressure was markedly attenuated. Serial studies of cardiac hemodynamics recorded before and after alleviation of tamponade were used as a guide to treatment of 3 of our patients as shown in Figures 3, 4, and 5. Figure 3 illustrates a transition from tamponade to normal intracardiac pressures and disappearance of the paradoxical pulse following removal of pericardial fluid by pericardicentesis alone. Figure 4 shows a right atrial pressure obtained prior to operative intervention and a modest reduction of the right atrial pressure following drainage of pericardial fluid through a subxiphoid pericardiotomy. Intrapericardial thrombus could be palpated and was evacuated through the median sternotomy, returning the intracardiac pressures to normal. In Figure 5, the elevation of the right ventricular enddiastolic pressure is evident; however, this value did not exceed one-third the peak systolic right ventricular pressure. The major hemodynamic abnormality was failure of the right ventricular pressure to return to baseline, indicating impairment of diastolic filling. A small volume (60 ml) of fluid was drained through a subxiphoid pericardiotomy; no thrombus was palpable, the early right ventricular end-diastolic pressure returned to

5 159 Hardesty et al: Delayed Postoperative Cardiac Tamponade I 1 I pre-per1 CARD~OCENTES~S 1 I I POST-PERICARDIOCENTES IS O&.,..,.._,...,,.,,... a...,.. L B Fig 3. Transition from tamponade (A) to normal intracardiac pressures (Bi and disappearance of the paradoxical pulse following pericardicentesis in a patient with delayed postoperative cardiac tamponade. baseline, and the procedure was terminated. Early operative intervention may have prevented progression to tamponade. Echocardiography (Fig 6) and radionuclide angiocardiography may disclose the presence of pericardial effusion or thrombus. In 1 patient, a right atrial angiogram and radionuclide an- giocardiography (Fig 7) demonstrated a large thrombus adjacent to the right atrium. Subsequent hemodynamic measurements excluded cardiac tamponade, making pericardiotomy unnecessary (Fig 8). Progression to tamponade did not occur. Alleviation of tamponade has been achieved by pericardicentesis (1/8), subxiphoid pericardiotomy (2/8), and repeat median sternotomy (5/8). No mortality was attributed directly to the tamponade, but 1 patient died as a consequence of many subsequent complications that were indirectly related to this problem.

6 160 The Annals of Thoracic Surgery Vol 26 No 2 August w I OmmHg 8 OmmHg d 0 mmhg A B Fig 4. An elevated ri<pht atrial pressure associated with delayed tamponade (A) was reduced only modestly by a subxiphoid pericardiotomy (B) and returned to normal after the median sternotomy was reopened and a thrombus evacuated (0. C, PRE-PER1 CARD IOTOflY 10 -, ' I 1 I100 MMHG A Fig 5. Elevation of right ventricular end-diastolic pressure occurring postoperatively in a patient suspected of having delayed tamponade. The end-diastolic pressure does not equal one-third the peak systolic right ventricular pressure (A). Evacuation of intrapericardial fluid (60 ml) returned the early right ventricular end-diastolic pressure to normal fb).

7 161 Hardesy et al: Delayed Postoperative Cardiac Tamponade Fig 6. Echocardiogram demonstrating pericardial effusion in a patient suspected of having delayed tamponade. A Fig 7. (A) Right atrial angiogram demonstrating widening of the space between the atrial border and the pericardium, which could represent either thrombus or effusion. (B) Radionuclide angiocardiogram delineating an abnormally wide space between the right border of the heart and the right lung of the same patient as in A. B

8 . 162 The Annals of Thoracic Surgery VoI 26 No 2 August _._. I.. RIGHT ATRIAL,A -I PULMDNARY ARTERY PRESSURE._.-_-. WEDGE PRESSURE 0 MMHG. I PULMONARY ARTERY PRESSURE MMHG Fig 8. Hemodynamic data excluding tamponade in a patient with pericardial effusion confirmed by echocardiogram (see Fig 6) and by right atrial angiogram and radionuclide angiocardiography (see Fig 7). Comment The challenge of delayed cardiac tamponade occurring after operation is to recognize it early. The complication is not anticipated because of its infrequent occurrence [2, 8, 111. Consequently, the low cardiac output or refractory congestive heart failure is initially attributed to impaired ventricular function. The normal heart tones [4, 81 and normal electrocardiographic voltage [l, 4, 51 do not arouse suspicion. The cardiac silhouette on a roentgenogram does not have the typical water-bottle appearance characteristic of an effusion; the cardiothoracic ratio may be normal [2,4], or a large cardiothoracic ratio may be ascribed to congestive heart failure. Kussmaul s sign occurs late [5]. A paradoxical pulse is not readily apparent and must be sought; thus, a prerequisite for its detection is suspicion of tamponade. Additionally, a paradoxical pulse is a late sign [l, 51 that can also occur in congestive heart

9 163 Hardesty et al: Delayed Postoperative Cardiac Tamponade failure [8, 91. Accordingly, of primary importance in establishing the diagnosis is awareness that delayed tamponade is suggested by refractory congestive heart failure or low cardiac output which is inconsistent with expectations based on the preoperative status and operative events. The extremely variable and unpredictable prothrombinopenia, which is not commensurate with the amount of warfarin administered, is an important clue-and frequently an early one. The association of anticoagulation (warfarin and heparin) [l-3, 5, 8-10] with delayed tamponade, and the lack of correlation between its occurrence and the degree of anticoagulation [l], is a common experience. An unexplained decline in the hemoglobin and hematocrit and an increase in the cardiothoracic ratio on roentgenogram herald intrapericardial bleeding. Experience at the University of Pittsburgh indicates that the erratic prothrombinopenia is generally accompanied by slow or intermittent intrapericardial bleeding, leading to the diagnosis of tamponade because of the resultant large effusion that eventually is recognized roentgenographically. A rapid accumulation presents as shock before an increase in the cardiothoracic ratio is apparent. The abnormal results of hepatic function tests suggest that this unpredictable response to warfarin reflects hepatic dysfunction. It is this erratic response, not the degree of hypoprothrombinemia, that is an early clue to latent tamponade. In the absence of anticoagulants, variable hypoprothrombinemia is not likely to be associated with tamponade. Rarely, tamponade may be seen with compression of the right ventricular outflow tract [51 or left atrium [12, 141 or with accumulation of pericardial fluid as a consequence of a prior pericardiotomy syndrome [61 or chyle [71. Noninvasive echocardiography and radionuclide angiocardiography are useful methods for detecting an effusion or thrombus within the pericardium. Confirmation of tamponade due to the effusion or a thrombus, and exclusion of poor ventricular function as the basis for low cardiac output, can be achieved by cardiac catheterization or with a Swan-Ganz balloon flotation catheter [13]. Restriction of car- diac filling is indicated by virtually equivalent pulmonary capillary wedge, pulmonary artery end-diastolic, right ventricular end-diastolic, and right atrial mean pressure. Additional criteria include a right ventricular end-diastolic pressure that is equal to or greater than onethird the peak systolic right ventricular pressure and attenuation of the y descent in the right atrial pressure contour. Cardiac catheterization with angiography provides the additional diagnostic advantage of localizing a thrombus while excluding the possibility of a major pulmonary embolus. Hemodynamic measurements similar to those in tamponade may occur with chronic constrictive pericarditis and restrictive cardiomyopathy, but neither is a differential consideration in the postoperative patient. A pulmonary embolus results in pulmonary capillary wedge or left atrial pressure that is disproportionately lower than the pulmonary artery diastolic pressure, right ventricular end-diastolic pressure, and right atrial mean pressure. The intracardiac pressures associated with a right ventricular infarction are similar to those occurring with a pulmonary embolus [13], but this is a rare event that may be recognized electrocardiographically as a posterior or inferior myocardial infarction. Pericardicentesis, subxiphoid pericardiotomy, and median stemotomy or thoracotomy constitute three operative approaches to decompressing the pericardial space. Defibrinated blood, serous fluid of the postpericardiotomy syndrome, and chyle can be removed by a subxiphoid pericardicentesis using a large needle. A thrombus must be extracted by subxiphoid pericardiotomy or by mediastinal exploration. The adequacy of decompression is readily assessed by measurement of intracardiac pressures, either by the Swan- Ganz balloon flotation catheter or by catheters left in the right atrium and ventricle following cardiac catheterization. A rise in mean arterial pressure, fall in right atrial pressure, return of the right ventricular end-diastolic pressure to normal, and normal y descent of the right atrial pressure contour indicate adequate decompression. Should pericardicentesis alone fail to achieve optimal return of intracardiac pressures

10 164 The Annals of Thoracic Surgery Vol 26 NO 2 August 1978 to normal, a subxiphoid pericardiotomy may succeed in breaking loculations and evacuating fluid and thrombus. This, too, may be inadequate if the thrombus is adherent or if it is located adjacent to the high right atrium and superior vena cava, to the left atrium posteriorly, or to the pulmonary artery and right ventricle anteriorly. Reopening the entire initial incision remains the only option in these circumstances. Reopening the incision as the initial approach obviates the need for monitoring intracardiac pressures after pericardicentesis and subxiphoid pericardiotomy, but it also represents a more extensive procedure that may be unnecessary. References 1. Ellison LH, Kirsh MD: Delayed mediastinal tamponade after open heart surgery. Chest 65:64, Engelman RM, Spencer FC, Reed GE, et al: Cardiac tamponade following open heart surgery. Circulation 41:Suppl 2:165, Fraser DG, Ullyot DJ: Mediastinal tamponade after open heart surgery. J Thorac Cardiovasc Surg 66:629, Garcia JM, Reyes E, Cheanvechai C, et al: Delayed cardiac tamponade following open heart surgery. Cleve Clin Q 41:103, Hill JD, Johnson DC, Miller GE Jr, et al: Latent mediastinal tamponade after open heart surgery. Arch Surg 99:808, McCabe JC, Engle MA, Ebert PA: Chronic pericardial effusion requiring pericardiectomy in the postpericardiotomy syndrome. J Thorac Cardiovasc Surg 67:814, McGoon DC: Discussion of Garcia et a1 [4] 8. Merrill W, Donahoo JS, Brawley RK, et al: Late cardiac tamponade: a potentially lethal complication of open heart surgery. J Thorac Cardiovasc Surg 72:929, Nelson RM, Jenson CB, Smoot WM: Pericardial tamponade following open heart surgery. J Thorac Cardiovasc Surg 58:510, Previtt TA, Rackley CE, Wilcox BR, et al: Cardiac tamponade as a late complication of open heart surgery. Am Heart J 76:139, Scott RA, Drew CE: Delayed pericardial effusion after cardiac surgery. Br Heart J 35:1304, Simpkin P, Brown H, Ersoz A, et al: Chronic left heart tamponade. J Thorac Cardiovasc Surg 65:531, Weeks KR, Chatterjee J, Black S, et al: Bedside hemodynamic monitoring. J Thorac Cardiovasc Surg 71:250, Yacoub MH, Cleland WP, Deal CW: Left atrial tamponade. Thorax 21:306, 1966