Double valve replacement and reconstruction of the intervalvular fibrous body in patients with active infective endocarditis

Abstract

OBJECTIVES

Destruction of the intervalvular fibrous body, though uncommon, occurs due to paravalvular abscess formation following active infective endocarditis. This warrants a highly complex operation involving radical surgical debridement of the intervalvular fibrous body, followed by double valve (aortic and mitral) replacement with patch reconstruction of the anterior mitral annulus, the left ventricular outflow tract and the left atrial roof. The objective of this study was to review the early and mid-term outcomes in patients undergoing this operation.

METHODS

A total of 25 patients underwent double valve replacement with reconstruction of the intervalvular fibrous body for extensive infective endocarditis between January 1999 and March 2012. The mean age was 64.3 ± 10.5 years. Most of the patients (60%) were in New York Heart Association Class III–IV, 12% and in cardiogenic shock. Associated comorbidities like acute renal insufficiency and cerebrovascular accidents were observed in 40 and 20% of patients, respectively. Twenty patients had previous heart valve surgeries. The logistic EuroSCORE predicted risk of mortality was 55.1 ± 22.9%.

RESULTS

Overall, 30-day mortality was 32%. Postoperative complications like low cardiac output, stroke and acute renal failure developed in 16, 28 and 56%, respectively. Thirty-two percent of patients required re-exploration for bleeding. Nine patients were alive at a mean follow-up of 406 days (0–8 years). The 2- and 5-year survivals were 37.0 ± 11.1 and 24.6 ± 12.5%, respectively.

CONCLUSIONS

Double valve replacement with reconstruction of the intervalvular fibrous body for infective endocarditis is a complex, technically challenging operation associated with high perioperative morbidity and mortality. Nevertheless, being the only option available for such complex disease, it should be performed in these patients who, otherwise, face 100% mortality.

Endocarditis, Aortic valve replacement, Mitral valve replacement

INTRODUCTION

Infective endocarditis is a life-threatening disease associated with a high mortality. The incidence ranges between 30 and 100 episodes per million patient-years [1]. Owing to the rising number of intracardiac interventions and device implantations in patients with increasing age, multiple comorbidities and compromised immunity, more complex intracardiac infections requiring complex reconstructive surgery have been encountered in the last decade. This can also be attributed to the rapid progress in surgical techniques developed to counter complicated situations [2]. One such situation is the involvement of the intervalvular fibrous body (IFB). This is a fibrous structure located between the lateral and medial fibrous trigones and connects the anterior mitral leaflet and parts of the left and non-coronary cusps to the fibrous skeleton of the heart. The dome of the left atrium is also attached to it, separating the anterior mitral leaflet from the aortic root. Destruction of the IFB, particularly due to abscess formation following infective endocarditis, can make surgical intervention extremely challenging. Radical debridement of all infected tissue followed by replacement of the aortic and mitral valves with reconstruction of the IFB is the only option available to treat these patients [3–5]. The reported mortality is high and ranges between 20 and 30% [3, 4, 6]. A conventional heart valve operation is not radical enough to prevent recurrent infection and to achieve a successful outcome. Although a few case reports have been published in the literature [7, 8], no large case series exist that focus solely on infective endocarditis involving the IFB. This is probably because the disease is so rare that it is not very commonly encountered even in very high-volume centres.

This retrospective study analyses the early and mid-term outcomes of patients with infective endocarditis who underwent double valve replacement with reconstruction of the IFB body.

PATIENTS AND METHODS

Study design

The study was approved by our Institutional Ethics Committee. Being a retrospective study, individual patient informed consent was waived.

A total of 1571 patients underwent valvular heart surgery for infective endocarditis at the Leipzig Heart Center between January 1999 and March 2012. Of these, 729 patients underwent aortic valve replacement (AVR), 487 mitral valve replacement (MVR) and 222 combined aortic and mitral valve replacement (DVR). Of the latter, 25 (1.6%) patients underwent DVR with reconstruction of the IFB and form the focus on the present study.

Clinical, operative and outcome data were prospectively collected in a computerized database. Our primary outcome was operative mortality, which was defined as death occurring within 30 days of the operation.

Surgical technique

Following a median sternotomy, the patient was connected to cardiopulmonary bypass. Bicaval cannulation was used in all patients. The native or prosthetic aortic valve was excised after a standard oblique aortotomy, which was then extended into the non-coronary aortic sinus and annulus until it reached the dome of the left atrium. The left atrial roof was then opened, extending posteriorly up to the level of the right pulmonary artery and anteriorly to the aortic annulus, which was then excised en bloc with the anterior leaflet of the mitral valve and the IFB. The extent of resection depended on the degree to which the neighbouring tissues were involved by the infectious process.

Excision of all infected valve tissue in native valve endocarditis, all prosthetic valve material and a thorough drainage and debridement of all abscess cavities form the cornerstone of this procedure. Following this, the new mechanical or biological mitral prosthesis was implanted such that it is secured to the posterior mitral annulus from the lateral to the medial fibrous trigone (Fig. 1A–D). The anterior part of the annulus between these trigones, i.e. IFB, was then reconstructed by sewing the base of a two-tongued triangular glutaraldehyde-fixed bovine pericardial patch to the sewing cuff of the mitral prosthesis (Fig. 2A and B). The posterior tongue was then sewn to the open margins of the roof of the left atrium (Fig. 3A), and the anterior tongue was sutured to the margins of the aortic wall, thus forming the new subaortic curtain and the posterior segment of the left ventricular outflow tract. Depending on the extent of aortic annular involvement, a mechanical or biological aortic prosthesis, or an aortic root conduit (mechanical, tailor-made biological conduit, stentless root or homograft), was secured to the aortic annulus and to the pericardial patch (Fig. 3C and D). In case of an AVR, the patch was used to close the right side of the aortotomy. When a homograft was implanted, a single patch was used to close the left atrial roof, and the IFB was reconstructed by sewing the anterior mitral leaflet of the homograft to the sewing cuff of the mitral prosthesis.

Figure 1:

(A and B) Pledgeted 2.0 polyester sutures inserted into the posterior mitral annulus from the left to right fibrous trigones (thin black arrows). Thick white arrows indicate the coronary buttons. (C and D) Mechanical mitral prosthesis fixed to the posterior mitral annulus from the left to right fibrous trigones.

Figure 2:

(A and B) Double pericardial patch (black arrows) sutured to the mitral valve prosthesis between the fibrous trigones. Dotted lines show the eventual contour of the pericardial patch.

Figure 3:

(A and B) Posterior tongue of the pericardial patch sutured to the left atrial roof (black arrow); pledgeted 2.0 polyester sutures inserted into the aortic annulus and the anterior tongue of the patch for AVR. (C and D) Implantation of the aortic root conduit (black arrow).

Follow-up

Follow-up was performed by personal contact, mailed questionnaire or by phone contact with patients and family members, with supplemental information supplied by family physicians and referring cardiologists. The closing interval for this study was between February 2012 and March 2012. The follow-up was 100% complete.

RESULTS

Demographic characteristics

Patient demographic characteristics and intraoperative details are depicted in Tables 1 and 2. The patients' age ranged between 37 and 79 years, three quarters of them being older than 60 years. All patients underwent urgent or emergent surgery. The main indications for surgery were development of paravalvular abscesses (80%), vegetations ≥10 mm (60%), cardiogenic shock (12%) and prosthetic valve endocarditis (80%) or a combination of these. The patients in this study were extremely frail and morbid, with mean logistic EuroSCORE of 55.1 ± 22.9%. The microorganisms could be isolated in 88% of the blood samples: Staphylococcus aureus in 3, Staphylococcus epidermidis in 5, Enterococcus faecalis in 8, Streptococci in 4 and other bacteria in 1. Three (12%) patients had culture-negative endocarditis. The focus of infection was known in the majority (84%) of patients: 5 had a pulmonary infection, 2 each suffered from a gastrointestinal tract infection, wounds and extremities and 1 each from the urinary tract and ear, nose and throat. Almost half the patients (48%) suffered septic embolization: 5 to the brain, 11 splenic and 4 each to the kidneys and extremities. All patients underwent DVR with reconstruction of the IFB with glutaraldehyde-treated bovine pericardium.

Table 1:

Distribution of preoperative variables

Preoperative variable
Age in years	64.3 ± 10.5
Male	21 (84)
Systemic arterial hypertension	18 (72)
Diabetes mellitus	10 (40)
Hyperlipoproteinaemia	12 (48)
Renal insufficiency
Non-dialysis dependent	7 (28)
Dialysis dependent	3 (12)
Peripheral arterial disease	3 (12)
Preoperative neurological events	5 (20)
NYHA Class
I	4 (16)
II	6 (24)
III	9 (36)
IV	6 (24)
Left ventricular ejection fraction (%)
>60	7 (28)
51–60	7 (28)
36–50	8 (32)
≤35	3 (12)
Cardiogenic shock	3 (12)
Preoperative ventilation	3 (12)
Coronary artery disease	9 (36)
Valve pathology
Aortic stenosis ≥Grade II	3 (12)
Aortic regurgitation ≥Grade II	11 (44)
Mitral stenosis ≥Grade II	2 (8)
Mitral regurgitation ≥Grade II	15 (60)
Tricuspid regurgitation ≥Grade II	8 (32)
Native valve endocarditis	5 (20)
Prosthetic valve endocarditis	20 (80)
Aortic valve endocarditis	21 (84)
Mitral valve endocarditis	20 (80)
Aortic and mitral valve endocarditis	16 (64)
Vegetation size (mm)
≥10	15 (60)
5–9	5 (20)
≤4	2 (8)
Paravalvular abscess	20 (80)
Previous cardiac surgery	20 (80)
Aortic valve replacement	11
Mitral valve replacement	2
Aortic and mitral valve replacement	6
Aortic and mitral valve repair	1
Associated coronary artery bypass graft	6
Timing of surgery
Urgent	11 (44)
Emergent	14 (56)
Log EuroSCORE %	55.1 ± 22.9

Continuous variables expressed as mean ± standard deviation. Percentages are shown in parentheses.

NYHA: New York Heart Association.

Table 2:

Intraoperative data

Intraoperative variable
Reconstruction of intervalvular fibrous body	25 (100)
Patch material
Bovine pericardium	25 (100)
Aortic valve replacement
Mechanical	6 (24)
Biological	19 (76)
Stented bioprosthesis	4 (16)
Stentless porcine root	10 (40)
Homograft	5 (20)
Mitral valve replacement
Mechanical	6 (24)
Biological	19 (76)
Associated procedures
Tricuspid valve repair	4 (16)
Aortic surgeries
Aortic root replacement	15 (60)
Ascending aorta replacement	3 (12)
Partial arch replacement	2 (8)
Coronary artery bypass graft surgery	8 (32)
Left atrial ablation	1 (4)
Aortic cross-clamp time	171 ± 42
Cardiopulmonary bypass time	271 ± 86
Type of cardioplegia
Crystalloid	7 (28)
Blood	18 (72)

Postoperative outcomes

Overall, 8 (32%) patients died within 30 days of surgery. Operative mortalities for various subsets are summarized in Table 3. The major cause of death in 3 patients was uncontrollable bleeding due to poor patch anchoring in infection-ridden and devitalized tissues. One of them died in the operating room. One patient developed severe intractable right ventricular failure following an intraoperative myocardial infarction due to thrombus formation in the right coronary artery after aortic root replacement. He could not be weaned from cardiopulmonary bypass and succumbed in the operating room. Three patients died due to multiorgan system failure, 2 following sepsis and 1 following mesenteric ischaemia. One patient died due to massive haemoptysis following iatrogenic lung injury due to a falsely positioned central venous catheter. The postoperative complications and outcomes are summarized in Table 4. All patients received intravenous antibiotics for a minimum period of 6 weeks. The antibiotics administered were on the basis of the blood and/or intraoperative valve tissue sensitivity reports. If reports were not available at the time of admission or surgery, empirical antibiotic therapy (Vancomycin, Gentamicin and Imipenem) was implemented. It was immediately deescalated, once the culture sensitivity reports were available. All survivors underwent transthoracic echocardiography before discharge. Two patients each had small aortic and mitral paravalvular leaks, which were conservatively managed. They were neither haemodynamically significant nor did they result in haemolysis.

Table 3:

Operative mortalities in various subsets

Variable	Death/patients, n	Rate (%)
Preoperative characteristics
Age in years
≤50	1/6	16.7
60–69	4/12	33.3
≥70	3/7	42.9
Renal insufficiency
None	5/15	33.3
Non-dialysis dependent	3/7	42.9
Dialysis dependent	0/3	0.0
NYHA Class
I, II	3/10	33.3
III, IV	5/15	33.3
Left ventricular ejection fraction (%)
>60	1/7	14.3
51–60	4/7	57.1
36–50	3/8	37.5
≤35	0/3	0.0
Cardiogenic shock	0/3	0.0
Coronary artery disease
No	3/16	18.8
Yes	5/9	55.6
Microorganisms
Staphylococcus aureus	2/3	66.6
Staphylococcus epidermidis	0/5	0.0
Enterococcus faecalis	4/8	50.0
Streptococcus species	1/4	25.0
Others	0/1	0.0
None	1/3	33.3
Septic embolization	4/12	33.3
Previous cardiac surgery
No	1/5	20.0
Yes	7/20	35.0
Aortic valve surgery	3/11	27.3
Mitral valve surgery	0/2	0.0
Double valve surgery	4/7	57.1
Time period from previous operation (years)
<2	5/10	50.0
>2	2/10	20.0
Paravalvular abscess
No	1/5	20.0
Yes	7/20	35.0
Timing of surgery
Urgent	3/11	27.3
Emergent	5/14	35.7
Operative factors
Type of valve implanted
Mechanical	2/6	33.3
Biological	6/19	31.6
Aortic root replacement
No	3/9	33.3
Yes	5/16	31.3
Coronary artery bypass graft
No	4/17	23.5
Yes	4/8	50.0
Type of cardioplegia
Crystalloid	4/7	57.1
Blood	4/18	22.2

NYHA: New York Heart Association.

Table 4:

Postoperative outcomes

Outcomes	n (%)
Thirty-day mortality	8 (32)
One-year mortality	13 (52)
Low output syndrome	4 (16)
Postoperative inotropes	19 (76)
Re-exploration for bleeding	8 (32)
Sepsis	5 (20)
Postoperative pacemaker implantation	5 (20)
Postoperative neurological complications
Stroke	2 (8)
Convulsions	3 (12)
Hypoxic brain injury	2 (8)
Postoperative dialysis	14 (56)
Reintubation	6 (24)
Tracheostomy	8 (32)
Ventilation time (h)	323 ± 386
Intensive care unit stay (days)	16.6 ± 23
Hospital stay (days)	24 ± 22

Continuous variables expressed as standard ± mean deviation.

The mean follow-up was 13.5 ± 25.5 months and ranged from 0 to 102 months. No patient was lost to follow-up. There were 8 deaths during the follow-up, 5 occurring within the first year after surgery and contributing to a 1-year mortality of 52% for this complex procedure. Two patients had cardiac deaths (1 due to re-endocarditis), 4 died of multiorgan system failure and 1 each due to a cerebrovascular accident and a tumour. The 2- and 5-year survival rates including operative mortality were 37.0 ± 11.1 and 24.6 ± 12.5% respectively.

Two patients required reoperation. One patient developed mitral prosthetic endocarditis on the 46th postoperative day. He underwent a complex reoperation (MVR, aortic root replacement and reconstruction of the IFB), but died 3 months later of mesenteric ischaemia following a difficult and prolonged postoperative course. The second patient developed a paravalvular leak around the aortic prosthesis 1.5 years postoperatively. It was successfully closed by repeat surgery. Freedom from reoperation at 5 years was 83.3 ± 15.2%.

DISCUSSION

The development of a paravalvular abscess is one of the most dreaded complications of infective endocarditis, warranting urgent surgery. It makes surgical correction technically demanding and complex as the supporting fibrous skeleton of the heart is most often affected and extensively devitalized by the infectious process. When this disease destroys the IFB, the problem is compounded. Fortunately, it is uncommon, occurring only in 1.6% of all patients who underwent double valve surgery for infective endocarditis in our institution. There is very little evidence on the optimal surgical management of an infected IFB in the literature [9, 10]. The Toronto group, which has developed the surgical techniques to manage this problem, operates on approximately 3–5 such patients annually out of a total annual case load of 650 valve procedures [3]. A team approach is required to achieve satisfactory postoperative outcomes. First, preoperative detection of an abscess involving the IFB by echocardiography or thoracic computed tomography scans is imperative to formulate an effective plan of treatment [11]. Secondly, the surgeon should be experienced enough to recognize and excise all infected and devitalized tissues resulting in a large defect that has to be perfectly reconstructed. Finally, excellent postoperative management in the intensive care unit cannot be overemphasized.

Our institution, being a tertiary referral centre, receives patients from surrounding referral hospitals, often when medical management has proven futile. This delay results in the deterioration of the patients' clinical condition, which is complicated by the development of renal insufficiency (40%), septic emboli (48%), congestive heart failure and cardiogenic shock (12%), some even requiring ventilation before they are transferred to our institution. Hence, most of the patients are extremely sick prior to surgery, which is evident from the high logistic EuroSCORE predicted risk of mortality of 55.1 ± 22.9% in this series.

No patient with preoperative dialysis-dependent renal failure, cardiogenic shock or left ventricular ejection fraction <35% died (Table 3). This was most likely due to the small number of patients presenting with these problems. Secondly, 2 of the 3 patients in cardiogenic shock were primary cases with no history of previous heart surgery, making the procedures less complex than reoperations. Similarly, 2 of the 3 patients with poor left ventricular function underwent non-emergent surgery, suggestive of a better preoperative clinical status.

It is well known that superior results can be achieved by early surgery for active endocarditis [12, 13]. This is particularly true for left-sided valve infections caused by S. aureus and other virulent organisms. In our series, almost two-thirds of the infections were caused by E. faecalis and Staphylococci, and they accounted for the majority of (75%) operative deaths (Table 3). Although David et al. [14] found S. aureus and epidermidis to be the most common offending microorganisms in native and prosthetic valve endocarditis, respectively, they did not affect operative mortality. However, a previous report from the same group stated that Staphylococcus species was an independent predictor of poor outcome [15]. These organisms are extremely virulent and can rapidly destroy valve tissue and penetrate the surrounding structures, evolving into a paravalvular abscess.

In our study, 80% of the patients had a paravalvular abscess, because only those patients in whom the IFB was involved in the infectious process were included. Complete resection of the abscess and surrounding inflamed structures followed by reconstruction of the IFB and MVR with aortic valve or root replacement is a formidable surgical procedure requiring long operative, cardiopulmonary bypass and ischaemic times (Table 2) and prolonged intensive care and hospital stay (Table 4). David and colleagues [5] have always stressed that total eradication of the infection was the cornerstone in achieving not only acceptable early outcomes, but also satisfactory long-term results.

The choice of prosthesis in patients with infective endocarditis has always been controversial. Although few centres propose the use of homografts for patients with paravalvular abscesses as a barrier against recurrent endocarditis [16, 17], there are many reports in the literature that recommend otherwise [18, 19]. Moon et al. [20] have also reported that survival was independent of valve type, and the long-term freedom from reoperation in patients older than 65 years was similar in all patients with mechanical valves. Although homografts are our preferred option in patients with isolated aortic root abscesses without IFB involvement, the choice of valve in patients requiring IFB reconstruction is the same as for any patient requiring valve replacement for non-infectious causes. In patients without paravalvular abscesses, the involvement of the IFB is first detected after exploring the aortic root. In such cases, we do not hesitate to use stentless xenografts that are immediately available. In this series, a fourth of patients received mechanical valves based on their age (Table 2), despite poor long-term prognosis observed in patients with this disease spectrum. Of these, 4 patients were very young. Young patients tend to do the better long term, as they seldom have pre-existing comorbidities and have a higher probability of recovering from acute injury to other organ systems caused by the septic process. Two older patients, due to severe adhesions and challenging anatomy, received mechanical valves with the idea of circumventing the possibility of another complex reoperation in the future.

Surgical correction of paravalvular abscesses involving only one valve is itself associated with a high mortality. David et al. [14] reported an operative mortality of 11% in patients undergoing surgery for either aortic or mitral annular abscess. Yankah et al. [17] studied 161 patients with aortic root abscess who underwent freehand aortic valve and aortic root replacement with an antibiotic-treated cryopreserved aortic homograft. Thirty-day mortality was 9.3% for urgent, and 14.3% for emergency surgery. In our series, not only were both left-sided valves involved, the IFB was also destroyed either by infection or by abscess formation. Overall operative mortality in our study was 32% (35 and 20% in patients with and without paravalvular abscess as shown in Table 3), while the 1-year mortality, which is considered the ‘true’ mortality for complex procedures by some experts, was 52%. The Toronto group has also reported an operative mortality between 20 and 30% in this group of patients [6, 14]. Siniawski et al. studied 108 patients suffering from root abscess, of whom 53 had secondary infective mitral valve disease requiring double valve surgery. The overall mortality rate was 19.4%; however, the group treated by double valve surgery had an early mortality of 26.4%. Double valve surgery was a strong predictor of early death on multivariate analysis. Contiguous spread of infection from the aortic to the mitral valve was also one of the independent predictors of death [21]. Leontyev et al. [22], likewise, published that patients with concomitant mitral valve endocarditis had higher mortality compared with those without (36.2 vs 20%, P = 0.04). This is in sharp contrast to the report on surgery for double valve endocarditis published by Gillinov et al. [23], in which there was no operative death in 54 patients treated at the Cleveland Clinic. In this series, however, endocarditis was active in only 50% of patients and remote/healed in the other half. Most of the patients (78%) underwent elective surgery, and no patient had contiguous spread from one valve to another or involvement of fibrous trigones. Only 11% patients had abscess cavities. Lastly, this study was confined to native valve endocarditis.

Eighty percent of our patients had prosthetic valve endocarditis after one or more previous heart valve operations. Operative mortality in this subset of patients was 35%, being the highest (57.1%) in those with previous double valve surgery (Table 3). These operations are technically the most demanding, and the patients themselves are extremely sick due to the early development of abscesses in the IFB. Half of the patients requiring reoperations within 2 years of previous surgery died, when compared with a 20% operative mortality in patients operated after 2 years (Table 3). This is because patients undergoing early reoperations have a compromised general health status, not having yet fully recovered from their previous operations, severe adhesions making reoperations technically challenging and multiple comorbidities contributing to unfavourable outcomes. Our results are comparable with those published by Musci et al., who evaluated 349 patients with prosthetic valve endocarditis. Two-thirds had a paravalvular abscess. In-hospital mortality approached 30% (28.4%) and was predicted by age, preoperative catecholamine use, mechanical support, mitral valve replacement and emergency surgery [24]. Our results are conflicting with those of Lopes et al., who retrospectively analysed 41 patients undergoing allograft aortic root replacement for prosthetic valve endocarditis. In-hospital mortality in the Lopes series was 4.8%. The patients were younger, with only 25% requiring emergent surgery for haemodynamic instability. Although 83% of patients had a periannular abscess, not all had active endocarditis and none required additional MVR or reconstruction of the IFB [25]. Even though such reoperations are challenging, their level of difficulty is much lower when compared with reoperations for double valve prosthetic endocarditis requiring redo DVR with reconstruction of the IFB.

A significant mortality was also seen in the mid term, resulting in a 5-year survival rate of 24.6 ± 12.5%. This was, obviously, driven by the high operative and 1-year mortality. Being a prolonged, highly complex operation does tend to have a negative impact on survival. These patients are at a high risk for recurrent endocarditis due to an increased possibility of intraoperative contamination (long operative times) and perioperative bacteraemia (prolonged requirement of central venous, pulmonary artery and dialysis catheters). Valve dehiscence, paravalvular leak and pseudoaneurysms are also very much a possibility due to extensive use of pericardial patches to anchor the prosthetic valves. The above-mentioned complications most commonly present within the first year after surgery.

CONCLUSION

Destruction of the IFB by infection or abscess is a deadly disease that warrants emergency surgery. These patients need a radical debridement of all infected and devitalized tissues followed by DVR with reconstruction of the IFB with a glutaraldehyde-preserved pericardial patch. Though this operation is associated with a very high early and mid-term morbidity and mortality even in experienced centres worldwide, we believe that it is the only option available for this small but ill-fated group of patients, who otherwise face 100% mortality. A conventional DVR cannot be successfully performed in these patients.

Conflict of interest: none declared.

ACKNOWLEDGEMENTS

The authors thank Christiane Luehr for the line diagrams.

Author notes

†

Both authors equally contributed to this work.