P have important clinical implications. In most series of

Transcription

1 Primary Cardiac Valve Tumors Fred H. Edwards, MD, Douglas Hale, MD, Amram Cohen, MD, Lenardo Thompson, MD, A. Thomas Pezzella, MD, and Renu Virmani, MD Department of Surgery, Walter Reed Army Medical Center, Washington, DC; Uniformed Services University of the Health Sciences, Bethesda, Maryland; and Armed Forces Institute of Pathology, Washington, DC To investigate the characteristics of primary cardiac valve tumors, we retrospectively analyzed our multiinstitutional experience from 1932 through We encountered 56 valvular tumors in 53 patients. The average age of these patients was 52 years (range, 2 to 88 years) and 79% (42153) were male. Symptoms were present in 38% (20/53) and were neurological in 15% (8/53). Four patients experienced sudden death. Each of the four valves.was affected with approximately equal frequency: 16 aortic, 15 mitral, 13 pulmonary, and 12 tricuspid. All but four tumors were benign. The most common histological type was papillary fibroelastoma (41), followed by myxomas (5), fibromas (4), sarcomas (2), hamartoma (l), hemangioma (l), histiocytoma (l), and undifferentiated (1). Average tumor size was 1.15 cm (range, 3 mm to 7 cm), and the average size of fibroelastomas was 8 mm (range, 3 to 15 mm). valve tumors were more likely than aortic valve tumors to produce serious neurological symptoms or sudden death (8/15 versus 3/16; p < 0.05). Six patients underwent echocardiography, and results were positive in each, All 6 underwent uncomplicated valve repair or replacement. Compared with a series of 407 nonvalvular tumors, cardiac valve tumors are more likely to occur in male patients (p < 0.001) and adults (p < 0.001). Valve tumors are also more commonly benign (p < 0.001) and asymptomatic (p < 0.001). These tumors demonstrate somewhat less aggressive behavior compared with nonvalvular tumors, but their distinct propensity to produce serious clinical sequelae argues in favor of surgical resection for all cardiac valve tumors. (Ann Tlzorac Surg ) rimary cardiac valve tumors are quite rare but can P have important clinical implications. In most series of cardiac tumors, primary valve tumors comprise less than 10% of cases [l-61. The majority of information regarding these tumors has accumulated from isolated case reports and from general reports of cardiac tumors. Consequently, only scattered reports, each consisting of a small number of valve tumors, are presently available. To specifically investigate the clinical characteristics of this entity, we reviewed our experience with 56 primary cardiac valve tumors from 1932 through Material and Methods The records from Walter Reed Army Medical Center, Brooke Army Medical Center, and the Armed Forces Institute of Pathology were reviewed for tumors primarily involving the cardiac valves. We excluded those tumors that were metastatic or infectious. Tumors extending into extravalvular structures were included only if the major part of the tumor clearly involved a cardiac valve. Information was obtained from patient records, autopsy reports, and direct histological examination of the excised tumors. Adequate records were available from 1932 through Accepted for publication June 7, 1991 The opinions and assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense. Address reprint requests to Dr Edwards, Cardiothoracic Surgery, Walter Reed Army Medical Center, Washington, DC Statistical calculations were made using a standard software program (BMDP Statistical Software, Los Angeles, CA). Univariate comparison of selected parameters was tested by calculating a 2 value with 2 x 2 contingency tables and one degree of freedom except where otherwise stated. A p value of less than 0.05 was considered significant. Results Fifty-three patients were found to have 56 primary cardiac valve tumors. Forty-two (79%) were male, and the age ranged from 2 years to 88 years, with an average of 52 years of age (Fig 1). Only 1 patient was less than 21 years of age. Presen fa f ion Most patients were asymptomatic. Twenty patients (37.7%) had cardiopulmonary or neurological symptoms and 4 had sudden death. Among these symptomatic patients, there were significantly more left-sided tumors than right-sided tumors ( p < 0.05). The mitral valve was most commonly associated with symptoms (lo), followed by the aortic valve (7), tricuspid valve (5), and pulmonary valve (4) ( p = not significant). Excluding the 4 patients with sudden death, most symptomatic patients (14/20) had complaints consistent with moderate congestive heart failure, whereas 40% (8/20) had neurological symptoms (Table 1). The majority of patients with heart failure had small tumors, although at least 3 had tumors of sufficient size to by The Society of Thoracic Surgeons /91/$3.50

2 1128 EDWARDS ET AL Ann Thorac Surg 20 MI tr 15 < >70 Age in Years Fig 1. Age distributioii of patients with cardiac valve tumors. seriously obstruct flow across the valve (see Table 1). Each of these three tumors was malignant, and each was addressed with operative resection. The remaining tumors associated with heart failure were incidentally discovered at autopsy, thereby making it impossible to Table 1. Clinical Presentation" Valve Tumor Presentation Involved Type Congestive heart failure Neurological Sudden death Sarcoma Undifferentiated MFH Hamartoma Undifferentiated Sarcoma MFH Hemangioma Fibroma a There were 20 patients having a total of 22 symptoms. MFH = malignant fibrous histiocytoma; = papillary fibroelastoma. P"lM> 13 I CIJsipld 12 FIX 2. Location of value tumors. Each of the four valves was invojved ziiith approxiiiiafely equal frequency. determine whether these tumors played a role in the development of symptoms. In each of these cases, no other cardiac abnormalities were noted. The 8 patients with neurological symptoms (see Table 1) made up 15% of this series (8/53) and 40% (8/20) of those who were symptomatic. Six patients had mitral valve tumors, half of which were myxomas and half of which were malignancies. All neurological symptoms were unequivocal, but only 1 patient suffered a stroke and only 1 had a transient ischemic attack. None of the neurological sequelae were fatal. Four patients had sudden death. A left-sided valvular tumor was present in each instance. Two had acute myocardial infarctions from papillary fibroelastomas of the aortic valve, which had embolized to the left anterior descending coronary artery. The other 2 had mitral valve tumors. One of these was a large hemangioma and the other was a fibroelastoma of unspecified size. In each of the mitral valve cases, no direct autopsy evidence was found to link the cardiac tumors with the cause of death. Findings on physical examination included clinical signs consistent with congestive heart failure in 14 patients and neurological findings in 3 (dysarthria, rig,htsided extremity weakness, and confusion). Only 2 patients had firmly documented new systolic murmurs. Tumor Characteristics The size of tumors ranged from 3 mm to 7 cm with an average of 1.15 cm in largest dimension. As shown in Figure 2, the aortic valve was most commonly involved. A 2 analysis using a 2 x 4 contingency table, however, revealed that there was no significant difference in the frequency of involvement for each1 of the valves. Fifty-two of the 56 tumors (92.8%) were benign. There were 41 papillary fibroelastomas, 5 myxomas, 4 fibromas, 1 hamartoma, and 1 hemangioma (Fig 3). All of the aortic valve tumors were benign, as were 13 of 15 mitral valve tumors and all but one of the pulmonic and tricuspid valve tumors.

3 Ann Thorac Surg EDWARDS ET AL 1129 Flbroeiastoma Fig 3. Histological distribution. \ 1 / Fibroma 4 6 Hemangloma 1 Hamartoma 1 Sarcoma 2 Other Cancer 2 The remaining four tumors were malignant. There was 1 malignant fibrous histiocytoma of the anterior mitral valve leaflet, 1 undifferentiated carcinoma of the tricuspid valve, and 2 sarcomas, 1 involving the pulmonic valve and the other involving the mitral valve. All malignant valve tumors were associated with preoperative symptoms, three of which were neurogenic. None of the malignancies were seen with extracardiac metastatic disease and none produced sudden death. Three of the 4 patients with malignant tumors were female, and there was a statistically significant difference in the frequency of cancerous valve tumors among female patients (27%) as compared with male patients (2%) (p < 0.01). Because of the small number of malignant tumors, it was not possible to draw conclusions regarding the possibility of age as a risk factor for malignancy. The oldest (88 years) as well as the youngest patient (2 years) in this series had malignant valve tumors. The other 2 patients with cancerous valve tumors were 31 and 63 years of age. Five patients had multiple cardiac tumors. Three of these involved two different valves and all were papillary fibroelastomas (pulmonic and aortic, tricuspid and mitral, tricuspid and pulmonic). The other 2 patients had multiple tumors involving extravalvular structures (aortic valve malignant fibrous histiocytoma and left atrial appendage malignant fibrous histiocytoma, tricuspid valve undifferentiated carcinoma and right ventricular free wall fibroma). Diagnostic Studies Valve tumors were discovered at autopsy in 45 of the 53 patients, so few diagnostic studies were performed in this series. In the 22 symptomatic patients, all studies were normal except in 8 patients. In these patients, echocardiography was performed in 6 and revealed a mitral valve tumor in each instance. In another, chest radiography demonstrated a lung abscess. Documentation on 1 other patient indicates only that preoperative evaluation revealed a tricuspid valve tumor. Treat men t One patient with a mitral valve sarcoma had multiple cerebral infarcts and did not undergo operative interven- tion. All others with a preoperative diagnosis of valvular tumors were surgically explored. Our general approach was to remove the tumor while preserving the native valve if at all possible. Four patients with mitral valve papillary fibroelastomas were treated with excision of the tumor and valve repair. One patient with a mitral malignant fibrous histiocytoma had a mitral valve replacement and another patient with an undifferentiated carcinoma had a tricuspid valve replacement. Both patients in whom valve replacement was necessary had large malignant tumors, which precluded valve repair. An additional patient had an open lung biopsy and open drainage of a lung abscess which resulted in his death. At autopsy the lung abscess was found to be due to a pulmonary infarct that had been created by tumor embolization from a pulmonic valve sarcoma. All other patients undergoing operation survived the procedure. Long-term follow-up of these patients is not available. Comment Because of the rarity of cardiac valve tumors, there is an incomplete understanding of their clinical characteristics. Today it is particularly important to have an appreciation for the behavior and surgical implications of valve tumors because these lesions are being recognized more frequently [5-7]. Most valve tumors will be incidental findings at autopsy, but recently there has been an unequivocal increase in the premortem diagnosis of cardiac tumors, including valve tumors [5-71. This has been most directly attributable to the technologic advancements and more frequent use of echocardiography [5, 7, 81. It appears that a preoperative noninvasive diagnosis of even small valve tumors can be made with good confidence [9], and this will almost certainly result in an increase in the number of patients being referred for evaluation of these echocardiographic findings. To gain a better understanding of the clinical and pathological nature of these tumors we have reviewed our multiinstitutional experience over the last 58 years. Papillary Fibroelastoma The great majority of valve tumors are papillary fibroelastomas, so it is appropriate to describe this tumor in some detail. These benign lesions are characteristically small avascular tumors derived from normal components of the endocardium, fibrous tissue, elastic fibers, and smooth muscle cells. The majority are valvular; in the Armed Forces Institute of Pathology series, only 13% (6/45) were nonvalvular [2]. They most frequently arise from the valvular endocardium and grossly resemble a sea anemone with multiple papillary fronds (Figs 4, 5). When the atrioventricular valve is involved, the tumor is most often on the atrial side of the valve, but when semilunar valve is involved, there is no predilection for one side or the other. Along the valve leaflet, the most common site of occurrence is in the midportion, well away from the free edge or the annulus. Histologically, one finds a central core of dense connective tissue that appears to be a direct extension of the

4 1130 EDWARDS ET AL Ann Thorac Surg lyy1;52: Fig 4. Gross appearance of ppillary fibroelastorria. This tnnior arose from the uentricular surface of the aortic vnluc. endocardium. Surrounding this central core is a layer of hyperplastic endocardia1 cells, loose connective tissue, and a mesh of elastic fibers (Fig 6). Early reports on the clinical behavior of these tumors Fig 6. Microscopic appearance of papillary fibroelastonin. The tunior consists of multiple papillnry fronds arranged on a stalk which blends into the substance of the valve. (Hematoxylin and eosin, xzo befclre 3% enlar~yeiiient.) indicated that they were not responsible for cardiac symptoms [Z, 3, 10, 111, but our findings as well as those of others has shown this not to be invariably true. In fact, the association between aortic valve fibroelastornas and serious coronary or cerebral emboli has been dearly established [2, 3, 4, 8, 10, 12-16]. Fig 5. A more detailed view of the tunior shouln in Figure 4. The fronds are particularly friable at the most distal aspect and ma!/ produce serious coronaty or cerebral embolic events. Comparison With Nonvaluular Cardiac Tumors To see how these tumors differed from the more comrnon extravalvular cardiac tumors, we compared our series against the 407 patients having nonvalvular solid cardiac tumors reported in the Armed Forces Institute of Pathology series [2]. The most common nonvalvular tumor was, of course, the myxoma, and the most common valvular tumor in our series was the papillary fibroelastoma. Other major differences are outlined in Table 2. In general, valve tumors are significantly more likely to occur in adults and in male patients. In addition, they are more likely to be benign, asymptomatic, and discovered incidentally at autopsy. There was no significant difference in the frequency of multiple tumors or the occurrence of sudden death.

5 Ann Thorac Surg EDWARDS ET AL 1131 Table 2. Comparison of Characteristics Variable Valvular Nonvalvular p Value Male 79.2% (42153) 57.1% (76/133)b <0.001 Pediatric 1.8% (1/53) 21.1% (86/407) <0.001 Malignancy 7.6% (4/53) 30.7% (125/407) <0.001 Asymptomatic 47.2% (25153) 14.3% (58/407) <0.001 Sudden death 7.6% (4/53) 9.6% (391407) NS Multiple 9.4% (5/53) 13.5% (551407) NS a From the Armed Forces Institute of Pathology series [2] of 407 patients with solid nonvalvular cardiac tumors. Where specifically stated (76 males/133 cases). NS = not significant (p > 0.05). CZinicuZ Implications Primary cardiac valve tumors are less likely to produce serious sequelae as compared with nonvalvular tumors. Nevertheless, these tumors have been associated with sudden death, embolic events, and heart failure in a number of well-documented cases [l, 10, 12-16]. Perhaps because of its typically small size, the papillary fibroelastoma has been considered an innocuous lesion in the past [2, 3, 10, 111, but these tumors have a significant propensity to embolize [9, 10, 12-16]. The embolic material may arise from either fragments of the tumor or thrombus that not uncommonly forms around the tumor [9, 121. The potential for life-threatening complications indicates that cardiac valve tumors should be managed surgically. This is particularly advisable given the ability to repair rather than replace many affected valves. Once the diagnosis of a valvular tumor is made, it appears reasonable to begin systemic anticoagulation and carry out a prompt surgical resection of the tumor. Limitations of the Study More than 70% of these tumors are papillary fibroelastomas, so that generalizations regarding valvular tumors are necessarily skewed to reflect the characteristics of fibroelastomas. Also, most of the tumors in this series were discovered at autopsy, so treatment recommendations are drawn from the management of a relatively small number of patients. In addition, the military population of this series may not represent a typical cross-section of the general population. Conclusions Cardiac valve tumors constitute a distinct clinical entity with important differences from extravalvular cardiac tumors. The valves are affected with approximately equal frequency, and compared with extravalvular tumors, they are more likely to be benign and asymptomatic. We have found that valve tumors are quite uncommon in the pediatric population and that male patients appear to have a higher propensity for the development of these tumors than do female patients. The most common of the valve tumors is the papillary fibroelastoma, which is characteristically a small tumor having a small but definite propensity to produce embolic sequelae. The clinical behavior of these tumors is somewhat less aggressive as compared with the more common extravalvular tumors. Nevertheless, there is considerable evidence that valvular tumors can cause life-threatening complications. Because of this, we believe that the diagnosis of a cardiac valvular tumor mandates prompt surgical resection of the tumor with preservation of the native valve. References 1. Chitwood WR. Cardiac neoplasms: current diagnosis, pathology, and therapy. J Cardiac Surg 1988;3: McAllister HA, Fenoglio JJ. Tumors of the cardiovascular system. In: Atlas of tumor pathology, 2nd series. Washington, DC: Armed Forces Institute of Pathology, McAllister HA. Primary tumors and cysts of the heart and pericardium. In: Harvey WP, ed. Current problems in cardiology. Vol 4 (2). Chicago: Year Book Medical, 1979: McFadden PM, Lacy JR. Intracardiac papillary fibroelastoma: an occult cause of embolic neurologic deficit. Ann Thorac Surg 1987;43: Dein JR, Frist WH, Stinson EB, et al. Primary cardiac neoplasms. Early and late results of surgical treatment in 42 patients. J Thorac Cardiovasc Surg 1987;93: Novick RJ, Dobell ARC. Tumors of the heart. In: Baue AE, Geha A, Hammond GL, Laks H, Naunheim KS, eds. Glenn s thoracic and cardiovascular surgery. Norwalk, CT: Appleton & Lange, 1991: Bortolotti U, Maraglino G, Rubino M, et al. Surgical excision of intracardiac myxomas: a 20-year follow-up. Ann Thorac Surg 1990;49: Shub C, Tajik AJ, Seward JB, et al. Cardiac papillary fibroelastomas. Two-dimensional echocardiographic recognition. Mayo Clin Proc 1981;56: Fowles RE, Miller DC, Egbert BM, et al. Systemic embolization from a mitral valve papillary endocardia1 fibroma detected by two-dimensional echocardiography. Am Heart J 1981; 102: Topol EJ, Biern RO, Reitz BA. Cardiac papillary fibroelastoma and stroke. Echocardiographic diagnosis and guide to excision. Am J Med 1986;80: Pomerance A. Papillary tumours of the heart valves. J Pathol Bacteriol 1961;81: Kasarskis EJ, OConnor W, Earle G. Embolic stroke from cardiac papillary fibroelastomas. Stroke 1988;19: Almagro UA, Perry LS, Choi H, Pintar K. Papillary fibroelastomas of the heart. Report of six cases. Arch Pathol Lab Med 1982;106:31& Campbell M, Carling W. Sudden death due to a fibrinous polyp of the aortic valve. Guy s Hosp Rep 1934;84: Harris LS, Adelson L. Fatal coronary embolism from a myxomatous polyp of the aortic valve: an unusual cause of sudden death. Am J Clin Pathol 1965;43: Butterworth JS, Poindexter CA. Papilloma of cusp of the aortic valve: report of a patient with sudden death. Circulation 1973;48:21> Hanson EC, Gill CC, Razavi M, Loop FD. The surgical treatment of atrial myxomas. Clinical experience and late results in 33 patients. J Cardiovasc Surg 1985;89: