ORIGINAL ARTICLES Heart Valve Operations in Patients With Active Infective Endocarditis Tirone E. David, MD, Joanne Bos, RN, George T. Christakis, MD, Paulo R. Brofman, MD, David Wong, MD, and Christopher M. Feindel, MD Division of Cardiovascular Surgery, Toronto Western Hospital and the University of Toronto, Toronto, Ontario, Canada Sixty-two consecutive patients underwent heart valve operation for active infective endocarditis. There were men and women whose mean age was 9 years (range, to 79 years). The infection was in the aortic valve in 7 patients, the mitral valve in 8, the aortic and mitral valves in, and the tricuspid valve in. Twenty-four patients had prosthetic valve endocarditis. Stuphylococcus and Streptococcus were responsible for 8% of the infections. Annular abscess was encountered in patients. Complex valve procedures involving reconstruction of the left ventricular inflow or outflow tract or both were performed in patients. There were three operative deaths (.8%). Predictors of operative mortality were prosthetic valve endocarditis, preoperative shock, and annular abscess. Patients were followed for month to months (mean follow-up, months). Only patient required reoperation for persistent infection. There were ten late deaths. Most survivors (9%) are currently in New York Heart Association class I or. The -year actuarial survival was 79% * 7%. These data demonstrate excellent results in patients with native valve endocarditis, and support the premise that patients with prosthetic valve endocarditis should have early surgical intervention. (Ann Thorac Surg 99;9:7-) nfective endocarditis of the heart valves is often cured I with antibiotics [I,. The characteristics of the offending microorganism and the location of the heart valve in which it is nested are probably the two principal determinants of the outcome of antibiotic therapy. Certain microbes do not respond to antibiotics alone, and the same microorganism can react differently depending on the site of the valve in which it is lodged. For instance, endocarditis caused by Staphylococcus aureus "an usually be cured with antibiotics alone when the native tricuspid valve is involved, but surgical intervention is frequently necessary when the native aortic valve is the site of infection. Antibiotics can be effective in eliminating a microorganism from a native valve but ineffective against the same microorganism in a prosthetic heart valve. Thus, although antibiotics are the principal treatment of infective endocarditis, operative intervention is an indispensable adjunct in many patients during the active phase of the infection. The difficulty is to determine when to intervene. As the results of valve surgery improve, a more aggressive approach to the management of these patients seems reasonable. This article summarizes our experience with patients who required a heart valve operation during the active phase of infective endocarditis. Material and Methods A review of our valve surgery registry indicated that from January 978 to June 989, patients underwent heart Presented at the Twenty-fifth Anniversary Meeting of The Society of Thoracic Surgeons, Baltimore, MD, Sep -, 989. Address reprint requests to Dr David, The Toronto Hospital, Elizabeth St, EN-9, Toronto, Ont, Canada MG C. valve operation during the active phase of infective endocarditis. Every patient was operated on during the course of antibiotic therapy or even before the diagnosis of infective endocarditis had been established. There were men and women whose mean age was 9 years (range, to 79 years). The principal indications for operation are shown in Table. Emergent operation was necessary in patients in septic or cardiogenic shock or both. In addition, symptoms of congestive heart failure were present in patients (7%), sepsis in (%), peripheral emboli in (%), cerebral emboli in 8 (%), and myocardial infarction in (%). The heart failure was severe in patients and moderate in 9. The preoperative electrocardiogram revealed sinus rhythm in 7 patients, atrial fibrillation in, and complete heart block in. Five patients had conduction disturbances ( with increased PR interval and with complete heart block) suggestive of annular abscess. Three other patients had evidence of recent myocardial infarction. Thirty-eight patients (%) had native valve endocarditis. Eleven of them had preexisting valvular lesions: aortic valve disease in (bicuspid in and rheumatic in ), mitral valve disease in (myxomatous in and rheumatic in l), and aortic and mitral valve disease in (rheumatic). Antecedent events thought to be responsible for the endocarditis were identified in patients: dental procedures in 8, remote infection in, and drug abuse in. Twenty-four patients (9%) had prosthetic valve endocarditis; 8 had mechanical valves and, tissue valves. Four patients had had composite replacement of the ascending aorta and aortic valve. The endocarditis occurred early (within months from the original operation) 99 by The Society of Thoracic Surgeons -97/9/$.
7 DAVID ET AL Ann Thorac Surg 99:9:7- Table. Indications for Operation, Incidence of Annular Abscess, Microorganisms, and Operative Survival Native Valves Prosthetic Valves Aortic + Aortic + Variable Aortic Mitral Mitral Tricuspid Aortic Mitral Mitral Total No. of patients Indications for operation Shock Heart failure Sepsis Emboli Annular abscess Microorganisms Staphylococcus aureus Staphylococcus epidermidis Streptococcus viridans Streptococcus faecalis Other streptococci Pseudomonas Other bacteria Undetermined Operative survival a Numbers in parentheses are percentages. () () () () () () 8 (9) 8 () () () () () (8) () 9 (9) in 9 patients and late in. Antecedent events were identified in 8 of the patients with late endocarditis: dental procedures in, drug abuse in, and remote infection in. Seventeen patients had had valve replacement once, twice, and three times before the operation for infective endocarditis. Blood cultures were positive in 8 patients and negative in. All surgical specimens were sent for culture, and microorganisms were isolated for 9 patients, including of the patients in whom blood cultures were negative. Thus, the microorganism responsible for the infection was identified in patients. Table shows the bacteria that caused endocarditis in these patients. Eight patients had received antibiotics for less than one day; the remaining patients had been on a regimen of antibiotics for one day to 7 days (mean duration, days) when operation was performed. Echocardiography was performed in all patients before operation. It revealed vegetations in the leaflets of 9 patients with native valve endocarditis and vegetations or abnormal masses in the leaflets of with a bioprosthesis. In addition, the diagnosis of aortic root abscess was made by echocardiography in 8 patients; had a fistula between the aortic root and surrounding structures, and had a paravalvular cavity suggestive of a false aneurysm. Eight patients underwent heart catheterization and coronary angiography before operation. An additional 9 patients with early prosthetic valve endocarditis had had coronary angiography before the original heart valve procedure. Major coronary artery disease was present in of these 7 patients. Operat ions The operations performed are summarized in Table. Patients with infection limited to the leaflets of the native or bioprosthetic valve were treated by valve replacement except for a few with tricuspid or mitral valve endocarditis in whom valve reconstruction was feasible. If the infection had extended into or beyond the annulus, aggressive debridement of all infected and inflamed tissues and reconstruction of the left ventricular inflow and outflow tracts and the atrial wall with either autologous or glutaraldehyde-preserved bovine pericardium was performed. When the abscess was limited to the aortic annulus, simple excision of that portion of the annulus and corresponding sinus was done. Reconstruction of the aortic root was performed with a properly tailored patch of pericardium secured to the healthy tissue around the defect by a continuous, running - polypropylene suture. When the abscess extended through the aortic wall into other tissues or cavities, a more extensive resection was performed. If the interventricular septum was involved, no serious consideration was given to conduction pathways, as the primary concern was to extirpate all necrotic and potentially infected tissues around the abscess cavity. In cases in which the abscess was in the ventricular muscle, the patch was secured to the left ventricular endocardium with a running - polypropylene suture. An aortic valve prosthesis was then secured to the aortic annulus in unresected areas and to the pericardial patch in areas where the annulus had been reconstructed.
Ann Thorac Surg 99;9:7- DAVID ET AL 7 Table. Operations Performed Operations No. of Patients Aortic valve endocarditis 7 () AVR b () RLVOT + AVR b () RLVOT + composite AVR and RAA () RLVOT + AVR + RAA () RLVOT i- AVR + RAA + MVR () RLVOT + composite AVR and RAA + MV repair () RLVOT + composite AVR and RAA + MVR () RLVOT AVR + MV repair () RLVOT + AVR + RRVOT + PVR + TVR Mitral valve endocarditis MVR () 8 () (7) MVR+RMA () MVR + AVR Aortic + mitral valve endocarditis AVR + MVR RLVOT + RMA + MVR + AVR Tricuspid valve endocarditis () TV repair () a Numbers in parentheses are percentages. Two patients also had aortocoronary bypass. One patient also had aortocoronary bypass. AVR = aortic valve replacement; Mv = mit~al valve; MVR = mitral valve replacement; PVR = p~lmo~ry valve replacement; RAA = replacement of ascending aorta; RLVOT = reconstruction of left ventricular outflow tract; Rh. = reconstruction of mitral annulus; RRVOT = reconstruction of right ventricular outflow tract; TV = tricuspid valve; TVR = tricuspid valve replacement. In a number of patients, the aortic root abscess extended into the fibrous skeleton that anchors the base of the anterior leaflet of the mitral valve, and in some patients, the inflammation had destroyed most of the anterior leaflet, thus making mitral valve replacement also necessary. In these patients, after the abscess was excised, the aortic and mitral annuli became a single orifice, and reconstruction was done with a triangular pericardial patch [,. Patients with multiple abscesses in the aortic annulus ended up with complete aortoventricular separation after debridement, and reconstruction of the left ventricular outflow tract was accomplished with a circumferential pericardial patch to cm high. One or both coronary arteries had to be reimplanted into this patch in several patients. Others required replacement of the proximal ascending aorta; this was done by suturing a composite graft with an aortic prosthesis directly to the pericardial patch []. Twelve of patients with native aortic valve endocarditis and of patients with prosthetic aortic valve endocarditis were found to have annular abscess at operation (see Table ). Patients with an abscess of the mitral annulus represented a greater challenge, and until 98, we simply debrided the abscess and secured the prosthetic valve in the surrounding atrial or ventricular muscle. We saw such patients, and the results were not satisfactory in either: died soon after operation, and in the other a recurrent infection developed; after reoperation, he was left with a leaking prosthetic mitral valve. A technique of mitral annulus reconstruction was then developed, and the results have been most gratifying [,. With this technique, the mitral annulus abscess is aggressively debrided, and the defect created is corrected with a properly tailored autologous or bovine pericardial patch []. The simplest situation is when a small segment of the posterior part of the mitral annulus is destroyed by infection. After appropriate debridement, a semicircular patch of pericardium measuring cm longer and cm wider than the defect in the annulus is prepared. The curved side of this patch is sutured to the endocardium of the left ventricle around the abscess cavity, starting and finishing at the normal fibrous annulus. We use a running - polypropylene suture, and the bites to the endocardium should be deeper in the trabeculated areas but never transmural, as they may damage major coronary arteries. The mitral valve prosthesis is secured to the healthy mitral annulus and to the straight side of the patch. More extensive destruction of the mitral annulus by multiple abscesses is better treated by circumferential reconstruction of the mitral annulus. A strip of pericardium is sutured to the endocardium of the left ventricle posteriorly and to fibrous skeleton superiorly. The mitral prosthesis is secured entirely in the pericardial patch [,. Simultaneous reconstructions of the aortic and mitral annuli are possible with a single patch sutured to the base of the left ventricle [-. Statistical Analyses Univariate analysis, with operative mortality as the end point, was performed on clinical, echocardiographic, bacteriological, and operative variables. A stepwise logistic regression analysis was applied to significant variables. Results There were three operative deaths, an overall operative mortality of.8%. All three deaths occurred in patients who had prosthetic valve endocarditis and were in shock when they underwent operation. Two deaths were due to low cardiac output syndrome after a prolonged operation in patients who required aortic and mitral valve replacement for multiple abscesses. The third death was due to hepatic failure in a cirrhotic patient who required aortic, tricuspid, and pulmonary valve replacement for multiple abscesses. Thus, the operative mortality for patients with prosthetic valve endocarditis was.%; there were no operative deaths among patients with native valve endocarditis. By univariate analysis, the following variables were predictive of operative mortality: prosthetic valve endocarditis, preoperative shock, and annular abscess. Stepwise logistic regression analysis indicated that only prosthetic valve endocarditis was significant. Postoperative complications were common. Nine patients required reexploration of the mediastinum for excessive bleeding or tamponade; no bleeding source was found in 8 patients, and it was attributed to coagulopathy.
7 DAVID ET AL Five patients required assisted ventilation for more than two days. Two patients sustained a perioperative stroke; recovered completely and, only partially. Low-output syndrome occurred in patients. Seven patients required dialysis for renal failure. Nine patients required a permanent transvenous pacemaker because of heart block. Two patients underwent amputation of a leg for control of sepsis, below and above the knee. Three patients had electrocardiographic evidence of myocardial infarction, but it was present before operation. All patients received a total of at least weeks of intravenous antibiotics. Patients in whom the surgical specimen was sterile received antibiotics for at least week after operation but no less than a total of weeks of preoperative and postoperative treatment. When the surgical specimen grew bacteria, antibiotics were given for postoperative weeks. Patients who had mechanical valves implanted were placed on a regimen of Coumadin (crystalline warfarin sodium) permanently. Patients with porcine valves received anticoagulants during the first postoperative months. Forty-nine patients (79%) are alive after a mean followup of +- months (range, to months). Thirtyeight patients (78%) are functionally in New York Heart Association class I; 9 (8%) are in class ; and only (%) are in class. There have been ten late deaths, five of which were cardiac related (only was valve related). Two patients were lost to follow-up at and months. One patient had recurrent mitral valve endocarditis because of mitral annulus abscess. He underwent reoperation, and was cured of his infection. However, perivalvular dehiscence developed, and he died in congestive heart failure years later. This happened before techniques for mitral annulus reconstruction [- were available. Two patients required valve rereplacement for structural valve deterioration of a porcine bioprosthesis (both mitral valves) after 8 years and years, and are currently asymptomatic. All patients have had a recent Doppler echocardiographic study. No patient has evidence of prosthetic valve dehiscence. One elderly woman has a mitral bioprosthesis with signs of degeneration, but she is not a candidate for reoperation. Six patients have sustained postoperative strokes ( with mechanical and with tissue valves). The actuarial survival is shown in Figure. The actuarial -year survival for all patients was 79% * 7%; for patients with native valve endocarditis, it was 8% * 8%, and for patients with prosthetic valve endocarditis, 7% * % (p = not significant). Comment Appropriate antibiotic therapy can eradicate the infection in most patients with native valve endocarditis [I,. Surgical intervention before completion of a standard course of antibiotics is reserved for patients in whom heart failure, persistent sepsis, recurrent emboli, annular abscess, or a combination of these develops [l]. The probability of operation being necessary during medical Ann Thorac Surg YY:Y:7- 'l-=--l t I- I Patients " 9 - Years Fig. Actuarial survival of patients with active infective endocarditis. treatment is greater in patients with left-sided endocarditis than those with right-sided endocarditis. On the left side of the heart, the aortic valve requires surgical intervention more often than the mitral valve [I,. Infections caused by certain microorganisms are also more likely to require surgical intervention because of their power to destroy tissue, eg, S aureus, or their resistance to antimicrobials, eg, Serratia, Pseudomonas, and Candida [I,, -8. There were no operative deaths among our patients with native valve endocarditis. These results are better than others previously reported [, 7, probably because the operations were performed during the past decade when improved methods of myocardial protection and better techniques to treat patients with annular abscess were available [-. In addition, all infections in our patients were caused by bacteria. The operative risk in patients infected with fungus is apparently high [l,, 7. Prosthetic valve endocarditis has always been associated with greater mortality and morbidity than native valve endocarditis [, 8. Antibiotics alone seldom eradicate infection in prosthetic valves [, 8-]. The pathological process of prosthetic valve endocarditis is different from that of native valves. Infection in a mechanical valve is always located in its sewing ring [ll]. In aortic valve homografts, the infection begins in the leaflets and destroys them very rapidly with consequent severe valvular dysfunction []. It seldom spreads into the annulus. Infections in porcine and pericardial bioprostheses behave differently depending on when they occur. Early infections involve the annulus or the leaflets or both. Late infections are frequently localized in the leaflets [lo]. Nufiez and associates reported on patients who had bioprosthetic valve endocarditis [lo]. Antibiotics alone cured 9 patients who did not have any echocardiographic evidence of valve dysfunction. Most of these infections were caused by streptococci. For the remaining patients, the valve was examined either at autopsy or operation, and the principal pathological findings were thickening of the leaflets, commissural fusion, and large, obstructing thrombotic vegetations. Only patients had annular abscess. Nine patients were treated surgically, and died. Our experience with prosthetic valve endocarditis is limited to patients, and although it is a small series for analyzing the significance of clinical and pathological variables, an operative mortality of.% (three deaths among patients) appears to be a remarkable improve-
Ann Thorac Surg 99;97- DAVID ET AL 7 ment in surgical results compared with results in previous reports [, 8-]. Our patients were comparable with those in other series: 8 were in shock when operated on, and 8 (7%) had annular abscess. Our improved surgical results reflect an increased experience with valve reoperation, better operative techniques to manage patients with extensive destruction of the left ventricular inflow and outflow tracts due to abscesses [-, improved methods to control coagulation disorders, which are frequently encountered in these patients [lo], and technological advances in the management of patients with multiorgan failure when they are transferred from the operating room to the intensive care unit. Our three operative deaths occurred in patients who were moribund and in shock before operation. We believe that patients with mechanical valve endocarditis should undergo operation as soon as the diagnosis has been established. Endocarditis in bioprostheses can be managed more conservatively providing there is no echocardiographic evidence of valve dysfunction [lo]. Regardless of the type of prosthetic valve, these patients should be hospitalized in institutions with open-heart surgical facilities, and a surgeon should be involved in the management of these patients from the time the diagnosis is being contemplated. This is probably the only way the mortality and morbidity will decrease in patients with prosthetic valve endocarditis. We are indebted to Dr A. Kerwin for his assistance in the preparation of the manuscript. References. Dinubile M. Surgery in active endocarditis. Ann Intern Med 98;9:-9.. Lowes JA, Hamer J, Williams G, et al. Ten years of infective endocarditis at St. Bartholomew s Hospital: analysis of clinical features and treatment in relation to prognosis and mortality. Lancet 98;:-.. David TE, Feindel CM. Reconstruction of the mitral annulus. Circulation 987;7(Suppl ):-7.. David TE, Feindel CM, Ropchan GV. Reconstruction of the left ventricle with autologous pericardium. J Thorac Cardiovasc Surg 987;97-.. David TE, Komeda M. Surgical treatment of aortic root abscess. Circulation 989;8O(Suppl ):9-7.. Richardson JV, Karp RB, Kirklin JW, Dismukes WE. Treatment of infective endocarditis: a -year comparative analysis. Circulation 978;8:8%97. 7. DAgostino RS, Miller DC, Stinson EB, et al. Valve replacement in patients with native valve endocarditis: what really determines operative outcome? Ann Thorac Surg 98;: 9-8. 8. Baumgartner WA, Miller DC, Reitz BA, et al. Surgical treatment of prosthetic valve endocarditis. Ann Thorac Surg 98;:87-. 9. Cowgill LD, Addonizio VP, Hopeman AR, Harken AH. A practical approach to prosthetic valve endocarditis. Ann Thorac Surg 987;:-7.. NuAez L, de la Llana R, Aguado MG, Iglesias A, Larrea JL, Calemin D. Bioprosthetic valve endocarditis: indicators for surgical intervention. Ann Thorac Surg 98;:-7.. Arnett EN, Roberts WC. Prosthetic valve endocarditis. Clinicopathologic analysis of necropsy patients with comparison of observation in 7 necropsy patients with active infective endocarditis involving natural left-sided cardiac valves. Am J Cardiol 97;8:8-9.. Clarkson PM, Barratt-Boyes BG. Bacterial endocarditis following homograft replacement of the aortic valve. Circulation 97;987-9. (For Discussion see page 7.)