Chapter 43. Surgical Treatment of Mitral Valve Endocarditis

Mitral valve infective endocarditis is one of the more devastating complications of heart valve disease, and, if left untreated, is universally fatal. Although the distribution of causes of mitral valve dysfunction has changed in recent years, the incidence of infective endocarditis has remained constant over the past several decades.1 Underlying rheumatic valvular disease, which was a frequent predisposing factor to infective endocarditis in the 1980s, is now rare in industrialized nations.2 Other predisposing factors more frequently encountered today include intravenous drug abuse, immunosuppression, degenerative valvular disease, intravascular prostheses and devices, hemodialysis catheters, and nosocomial infections. Several of these predisposing factors are consequences of advances that characterize modern medicine.3

Today, more effective antimicrobial agents have improved early and long-term outcomes of patients treated for endocarditis. However, endocarditis is still associated with high rates of morbidity and mortality and frequently requires operation for cure.4 Increased experience and improvements in surgical technique have improved success rates with these challenging operations.

Native Mitral Valve Endocarditis

Native valve endocarditis (NVE) refers to infectious endocarditis involving a patient's own (native) heart valve. Crude incidence of NVE is 6.2 per 100,000 people per year, and is highest in older age groups.5 The pathogenesis of NVE begins with endocardial trauma resulting in alteration of the valvular endocardial surface; this allows deposition of fibrin and platelets with subsequent attachment of bacteria. Endocardial injury may be secondary to rheumatic valvulitis or other leaflet disease, or valvular or annular calcification.5 Common reasons for bacteremia or fungemia predisposing to NVE include use of long-term indwelling catheters, intravenous drug abuse, and fungemia associated with prolonged antibiotic therapy.6,7 Although vegetations may be seen anywhere on the leaflets or the chordae, the usual site at which infective NVE of the mitral valve causes valvular destruction and invasion is at the base of the atrial aspect of the mitral valve leaflets. Annular or subannular invasion may cause separation at the atrioventricular (AV) junction. Fortunately, mitral annular invasion is shallow in most cases. Invasion into the AV groove fat with abscess formation is more serious and necessitates radical debridement. Destruction of the fibrous trigones and intervalvular fibrosa between the anterior mitral valve leaflet and the aorta is usually a consequence of aortic valve endocarditis with secondary mitral involvement. Occasionally “drop lesions” from an infected aortic valve seed the anterior mitral leaflet or the tensor apparatus of the mitral valve, resulting in double-valve endocarditis; the mechanism may be a large vegetation directly infecting the mitral leaflet or a jet lesion that becomes infected.

Prosthetic Mitral Valve Endocarditis

Prosthetic valve endocarditis (PVE) or replacement device endocarditis is infectious endocarditis involving a surgically implanted heart valve. The number of cases of PVE is on the rise as the number of patients with prosthetic heart valves continues to increase. In contradistinction to NVE, wherein the mitral valve is more likely to be infected,8 PVE is more common in the aortic than in the mitral position.9 The risk of PVE appears to be greatest approximately 5 weeks after valve implantation and thereafter declines.10–13

PVE identified within the first postoperative year is considered early endocarditis, and those cases appearing more than 1 year after operation are termed late.13,14 The incidence of early PVE is 1%.14 Once past the early phase, the incidence of late PVE is 0.5 to 1% per year.15–18 The type of prosthesis (mechanical versus bioprosthetic) does not influence the risk of PVE.

Early PVE is usually the result of intraoperative infection.19 Common portals of entry for bacteria that cause PVE are intravascular catheters and skin infection.20,21 Nosocomial infections contribute to late PVE, particularly in patients with medical comorbidities that require frequent hospital admission or instrumentation (eg, hemodialysis patients) or immunosuppression (eg, organ transplantation).19

Early PVE usually affects the sewing ring or the interface of the prosthetic valve and the annulus (often a site of clot formation), resulting in periprosthetic leak. Involvement of the sewing ring starts locally but eventually becomes circumferential. Enzymatic degradation of the tissue holding the sutures results in dehiscence of the prosthesis and periprosthetic leak. Progression of the infection and invasion may lead to abscess formation. Mitral PVE may extend anteriorly to the fibrous trigone or posteriorly, causing AV separation. PVE may be classified into different subgroups based on the anatomic distribution of the infection. According to this classification, PVE cases are classified as those involving the prosthesis alone, those involving the prosthesis–native annular junction (annular infection), and those extending beyond the annulus (extensive infection).13,19 Another useful classification that applies to both NVE and PVE involves the distinction between active and healed endocarditis. The latter includes cases in which organisms cannot be demonstrated but remote infection is still presumed.13,22

Endocarditis of native valves is most often caused by Streptococcus viridans, Staphylococcus aureus or epidermidis, or Enterococci. The distribution of microorganisms responsible for early PVE includes coagulase-negative staphylococci (52%), S. aureus (10%), Enterococci (8%), S. viridans (5%), and gram-negative organisms (6%).19 Fungi account for 10% of cases of early PVE (Candida albicans in 8 of 10). Although gram-positive cocci are dominant in both early and late PVE, in early PVE staphylococci predominate whereas in late PVE streptococci are the most common organisms isolated.9Haemophilus spp., Actinobacillus actinomycetemcomitans, Cardiobacterium hominis, Eikenella corrodens, and Kingella kingae are collectively termed the HACEK group and account for 3% of culture-negative late PVE.23 Patients with PVE and negative cultures may be infected by fastidious organisms. In general, late PVE is more amenable to successful antibiotic therapy than is early PVE; however, once the sewing ring is involved, cure is unlikely. PVE always warrants surgical consideration.

Clinical Presentation

The most common clinical finding is fever;9 however, it is nonspecific. Other findings include a new murmur or a change in an existing murmur. Embolic phenomena may cause petechiae, Roth spots, Osler nodes, and Janeway lesions. Splenomegaly may be present in both NVE and PVE.

Laboratory findings include a white blood cell count >12,000/mm3, anemia (hematocrit <34%), and hematuria.9 Blood cultures from separate sites are usually positive in patients with bacterial endocarditis; two of three positive cultures is considered diagnostic. However, cultures in patients with fastidious organisms or fungi may take more than 3 weeks to become positive. Blood cultures should be drawn before antibiotics are started.

Electrocardiographic Findings

Conduction abnormalities are present in up to 23% of patients with NVE and 47% of patients with PVE. This usually indicates extension of the infection and formation of a paravalvular abscess.24

Echocardiographic Findings

The present gold standard diagnostic modality for documenting infective endocarditis is transesophageal echocardiography. Specificity for transesophageal echocardiography is approximately 90% and sensitivity is 95%.26,27 In contrast, transthoracic echocardiography is more operator dependent and its images may be compromised by surrounding structures; it is only 50% sensitive and 90% specific for infective endocarditis.25 Echocardiographic findings of endocarditis include vegetations, periprosthetic leak in patients with PVE, intracardiac fistulae, and abscesses. The echocardiographic examination is very good at evaluating valve function but less reliable for assessing the severity and invasiveness of the infection. A negative echocardiogram does not exclude the diagnosis of endocarditis. In situations with strong suspicion of endocarditis, the diagnosis may be pursued by magnetic resonance imaging (MRI). In the majority of patients with endocarditis, MRI will demonstrate abnormal consistency of tissue in the annulus. Metastatic infection is a possibility and patients with infective endocarditis and abdominal symptoms should be investigated with computed tomography (CT) scanning to rule out splenic or hepatic abscesses. Metastatic infection of viscera is typically caused by staphylococcal organisms.5 The brain is the most frequent site for emboli.28,29 Any neurologic deficit or abnormality should trigger investigation with CT or MRI of the brain, funduscopic examination, and occasionally cerebrospinal fluid examination. CT or MRI of the brain is also justified in the absence of symptoms if there are stigmata of other embolic events. Preoperative coronary angiography is indicated in patients with coronary artery disease or history of previous revascularization procedures and in patients in whom coronary embolization is suspected.

Duke Criteria

The Duke University criteria were developed to improve the specificity and sensitivity of the diagnosis of infective endocarditis. These criteria include echocardiographic results along with clinical, microbiologic, and pathologic findings.30 Most authorities accept a modification of these criteria for PVE that includes progressive heart failure in the presence of positive blood cultures and new conduction disturbances (Table 43-1). The Duke criteria are classified as “major criteria” (typical positive blood culture and positive echocardiogram) and “minor” criteria (predisposition, fever, vascular phenomena, immunologic phenomena, suggestive echocardiogram, suggestive microbiologic findings, and new-onset heart failure or conduction disturbances). There are three diagnostic categories: (1) patients with a low clinical likelihood of infective endocarditis are those with resolution of manifestations of endocarditis with antibiotic therapy for 4 days or less, or no pathologic evidence of infective endocarditis at surgery or autopsy after antibiotic therapy for 4 days or less; (2) patients with possible endocarditis have one or two minor clinical criteria of the Duke classification system; and (3) patients with a definite diagnosis of infective endocarditis are those (a) in whom pathologic specimens from surgery or autopsy reveal positive histology and/or culture, (b) who have two major criteria, (c) who have one major and three minor criteria, or (d) who have five minor criteria.30 Understandably, the Duke criteria are not particularly relevant in patients who have had surgical treatment, those in whom the diagnosis is based on surgical findings, and those with analysis of surgical specimens by microscopy, culture, and polymerase chain reaction.

Surgery plays a pivotal role in the management of native mitral valve endocarditis. Indications for surgical intervention in patients with NVE and PVE are presented in Table 43-2. Congestive heart failure is the most common indication for surgery.

The majority of patients with PVE require surgery.31,32 Indications for surgical intervention in patients with PVE include heart failure, new heart block (possibly secondary to a myocardial abscess), ongoing sepsis, valve dehiscence, recurrent systemic embolism, relapse of infection, and fungal infection. Operation for large mobile vegetations to prevent embolization in both NVE and PVE remains a controversial indication in the absence of hemodynamic compromise or other surgical indications. Fungal PVE is uncommon but is more difficult to cure than is PVE caused by other organisms. The operative strategy should include perioperative intravenous amphotericin B, radical debridement of infected tissue, reconstruction using biologic tissue whenever possible, and lifelong oral suppressive antifungal therapy after completion of intravenous therapy.33

Many patients with NVE and the majority of patients with PVE require surgical intervention, and the challenge for the surgeon is to determine the appropriate timing of surgery.31 In most cases, the operation should not be delayed once surgical indications are present. Patients with PVE caused by S. aureus, even without periannular abscess formation, should have early operation to prevent an aggressive infection with rapid progression and invasion. The same principle applies to fungal infective endocarditis, because a mortality rate as high as 93% has been reported.9

However, delaying repair is usually advised in the presence of central nervous system complications.28 Before undergoing valve surgery, patients with endocarditis should have careful neurologic evaluation and significant findings should be evaluated by head CT or MRI. Occasionally, angiography is required to exclude a mycotic aneurysm. Patients with hemorrhagic strokes should have surgery delayed for 4 weeks to reduce the risk of further intracranial bleeding during heart surgery. Intracranial hemorrhage in the setting of PVE has been associated with mortality as high as 28 to 69%.35 The main concern with nonhemorrhagic embolic stroke is transformation of an ischemic infarct into a hemorrhagic infarct as a complication of the anticoagulation required during cardiopulmonary bypass.34 Patients who suffered recent ischemic stroke should have their operation delayed for 2 to 4 weeks, particularly if the stroke is large.36,37 The risk of worsening neurologic symptoms as a consequence of operation is time-related, decreasing with increasing interval from the initial neurologic event. This risk must be weighed against the indications for surgery and the risk of additional emboli during the waiting period.

General Principles

Operations for endocarditis are guided by some basic principles: optimal timing of surgery as discussed above, good exposure of the valve, radical debridement, optimal choice for reconstruction of the heart and repair or replacement of the valve, and adequate postoperative antibiotic treatment. Radical debridement with removal of all infected and necrotic tissue and foreign material is more difficult to accomplish in mitral cases with AV groove invasion and abscess formation, particularly when compared with aortic root infections. In addition, reconstruction after invasion of the AV groove and AV separation entails closing off the infected space, leaving it without drainage.

Native Mitral Valve Endocarditis

Intraoperative transesophageal echocardiography should be performed in all cases to evaluate the valve before commencing the procedure. Surgical treatment options for NVE affecting the mitral valve include valve replacement and valve repair. Although there is some experience with the mitral valve allograft for treatment of mitral valve endocarditis, there are too few data available to support this strategy.38

Most operations for NVE are best conducted through a full median sternotomy. Cannulation for cardiopulmonary bypass involves arterial return via the ascending aorta and bicaval cannulation for venous return. In case of large mobile vegetations, it is advisable to arrest the heart before placing a transatrial retrograde cardioplegia catheter. Protection is achieved using antegrade and retrograde substrate-enhanced cardioplegia.39

The mitral valve is exposed via a left atriotomy through the interatrial groove or transseptally. If the left atrium is small, an extended transseptal approach is employed. Once exposure of the mitral valve is accomplished, the valve is evaluated to assess for presence of paravalvular abscesses, intracardiac fistulae, or intervalvular fibrous body/ventricular involvement. Radical resection of all necrotic tissue is performed with a margin of normal tissue. All grossly infected tissue is removed without concern for the possibility of repair. Specimens are sent for microbiologic analysis and culture.

For NVE, our approach is to attempt repair if feasible. Mitral valve repair can be performed safely provided there is sufficient remaining tissue to allow valvular reconstruction without tension.40–44 In the event of extensive destruction of the subvalvular apparatus, prosthetic valve replacement is performed. Regardless of the mitral procedure performed, all patients with active infection receive 6 weeks of postoperative antibiotic therapy.

Anterior Leaflet Repair

“Drop lesions” of the anterior leaflet encountered in association with aortic endocarditis can be repaired with autologous or glutaraldehyde-preserved pericardium.38 A patch of pericardium is fixed to the remaining tissue of the anterior leaflet with running polypropylene suture (Fig. 43-1). The smooth surface of the patch should face the atrium to decrease the risk of thromboembolic complications.37 More extensive destruction involving both the aortic valve and the anterior leaflet of the mitral valve can be repaired using a free-standing aortic root homograft with the anterior leaflet of the mitral valve still attached. The homograft's attached aortomitral curtain can be used to reconstruct the base of the native anterior mitral leaflet.37,45

Involvement of the free margin of the anterior leaflet can be managed with triangular resection and closure with interrupted fine sutures. Anterior leaflet chordal rupture can be repaired with chordal transposition from the posterior leaflet or with secondary chordal transfer to the free margin of the anterior leaflet. Artificial chordae may also be used to replace ruptured anterior leaflet chordae.

Posterior Leaflet Repair

The middle scallop (P2 segment) of the posterior segment is frequently affected by the infectious process. Repair can be performed with quadrangular resection of the middle scallop, and a sliding repair is frequently required to close the gap between the remaining two scallops (Fig. 43-2). Extensive destruction of the posterior annulus requires removal of all devitalized tissue and annular reconstruction with autologous pericardium. Occasionally, repair of the annulus and the posterior leaflet can be accomplished with the same patch if chordal support to the leaflet is good. A running polypropylene suture is used on the ventricular, atrial, and valvular aspects of the patch. If replacement is required, a mechanical or bioprosthetic valve may then be inserted, affixing the prosthesis to the patch.45 It is important that the patch is generous enough to minimize tension on the sutures in ventricular muscle.

The use of prosthetic ring annuloplasty in NVE is controversial. Favoring repair over replacement, we use a ring whenever the annulus is not infected.

Prosthetic Valve Endocarditis

Surgical Approach

Operations for PVE are reoperations, usually performed via a median sternotomy. An alternative approach is right anterolateral thoracotomy in the fourth intercostal space; this is particularly useful in patients with multiple previous sternotomies, bypass grafts near the sternum, or a history of mediastinal radiation and/or mediastinitis.46,47 However, a right thoracotomy for mitral reoperations allows limited access, sometimes denies aortic cross-clamping, and may be associated with a risk of stroke.

Myocardial Protection

Cardiopulmonary bypass is instituted using the ascending aorta and bicaval cannulation. A transatrial retrograde cardioplegia cannula is used and myocardial protection is achieved using antegrade and retrograde cardioplegia.

Mitral Valve Exposure

The main issues in achieving mitral valve exposure are related to obtaining mobility of the right atrium and vena cavae.19 Our usual approach to obtain mitral valve exposure is to use an extended transseptal approach. If the left atrium is large and adhesions modest, a standard left atriotomy may be employed. Exposure may be enhanced by division of the superior vena cava and extending the left atriotomy toward the aortic root. This approach provides good exposure even when the left atrium is small.

Reconstruction of the Mitral Annulus

Once exposure of the mitral valve is obtained, the infected prosthesis is removed. Mitral valve PVE may produce an abscess cavity separating the left atrium, left ventricle, and prosthesis. In these situations the operation includes debridement of the annulus with subsequent annulus reconstruction using autologous or glutaraldehyde-fixed bovine pericardium (David technique).48,49 With this technique, a semicircular pericardial patch is used to reconstruct the annulus with one side of the patch sutured to the endocardium of the left ventricle and the other side to the left atrium. This patch closes off the cavity, which must be thoroughly debrided and sterilized before the patch is affixed. The new valve prosthesis is affixed to this reconstructed annulus (Fig. 43-3). The patch should be large in order to minimize tension on suture lines. In most situations with annular reconstruction we employ a bioprosthesis because of the larger and softer sewing ring and to avoid anticoagulation in the postoperative period.

An alternative technique for mitral annular reconstruction is the technique described by Carpentier and colleagues.50 This technique involves using figure-of-eight atrial and ventricular sutures to reconstruct the AV junction. Exerting traction on these sutures reduces the size of the annulus and closes the AV groove without injury to the circumflex vessels. The main potential disadvantage of this technique is that sutures may pull through stiff and noncompliant ventricular tissue.19 We have observed pseudoaneurysm formation after application of this technique.

Reconstruction of the Fibrous Trigones

Extension of PVE into the intervalular fibrosa/fibrous trigones may necessitate replacement of both mitral and aortic valves. This usually occurs in the setting of PVE affecting both the aortic and the mitral valves and seldom with isolated mitral valve endocarditis. Reconstruction of the intervalvular fibrosa as well as replacement of both the aortic and mitral valve are required (Fig. 43-4). In such circumstances the fibrous trigones may be reconstructed with autologous or bovine pericardium that is used to secure the new prosthesis.13,19 Perfect exposure is mandatory whether it is provided by the extended transseptal approach or by dividing the superior vena cava and extending the left atriotomy from anterior to the right superior pulmonary vein toward the dome of the left atrium. This approach allows debridement of the aortic and mitral valves, as well as the fibrous trigones. The prosthetic mitral valve is then sewn to the annulus posteriorly, medially, and laterally, and the superior portion of the mitral valve annulus is reconstructed with a pericardial patch that replaces the fibrous trigones. The valve is then sewn to the patch with horizontal mattress sutures. Once the mitral valve is secured in place, the aortic valve prosthesis is affixed to the aortic annulus. The pericardial patch is used to reconstruct the medial part of the aortic valve annulus. The aortic valve is then sewn to that patch.13,19 An alternative option is aortic valve and root allograft replacement in an anatomic position and orientation, suturing the intervalvular fibrosa/mitral valve of the allograft to the mitral valve prosthesis.

Native Mitral Valve Endocarditis

In patients with NVE, mitral valve repair is preferable to mitral valve replacement, because repair has been associated with lower hospital mortality and improved long-term survival (Fig. 43-5).43,51 In our series of 146 patients having surgery for NVE, patients undergoing repair had a lower hospital mortality (p = 0.008) and improved long-term survival (p = 0.05) compared with those having replacement.43 Several recent series have reported excellent results with valve repair in the setting of mitral valve endocarditis, with mortality rates ranging between 0 and 9%.40–43,52 The infection-free survival is also better for mitral valve repair compared with replacement with a less than 1% per year reinfection rate.43 The likely explanations for these excellent outcomes are the avoidance of prosthetic material in the infected field and the preservation of left ventricular function associated with repairing mitral valves.43 Second, the replacement group contains the sickest patients with the most advanced and destructive disease.

Prosthetic Valve Endocarditis

PVE is associated with a much higher operative mortality than is native valve endocarditis.24,49,53,54 Despite improved antimicrobial regimens, outcomes of medical therapy without reoperation are poor in PVE, particularly for patients with annular involvement or early endocarditis after valve surgery.13 The success of allografts in the aortic position has led to attempts to use mitral valve allografts to treat mitral valve PVE. That strategy seems to offer no advantage over standard prostheses and remains experimental.

For all patients with mitral valve PVE, prolonged postoperative treatment with intravenous antibiotics is required. Most patients with active endocarditis are treated for at least 6 weeks after surgery and are followed with transesophageal echo studies. In comparison with aortic valve endocarditis, radical debridement and drainage of posterior mitral annulus abscesses is far more difficult. Patients with fungal endocarditis require 2 months of intravenous antifungal therapy followed by indefinite oral antifungal therapy.33