Five-year outcomes following complex reconstructive surgery for infective endocarditis involving the intervalvular fibrous body

Abstract

OBJECTIVES

Destruction of the intervalvular fibrous body (IFB) due to infective endocarditis (IE) warrants a complex operation involving radical debridement of all infected tissue, followed by double valve replacement (aortic and mitral valve replacement) with patch reconstruction of the IFB. This study assesses the 5-year outcomes in patients undergoing this complex procedure for treatment of double valve IE with IFB involvement.

METHODS

A total of 127 consecutive patients underwent double valve replacement with reconstruction of the IFB for active complex IE between January 1999 and December 2018. Primary outcomes were 3-year and 5-year survival, as well as 5-year freedom from reoperation.

RESULTS

Patients’ mean age was 65.3 ± 12.9 years. Preoperative cardiogenic shock and sepsis were present in 17.3% and 18.9%, respectively. The majority of patients (81.3%) had undergone previous cardiac surgery. Overall, 30-day and 90-day mortality rates were 28.3% and 37.0%, respectively. The 3- and 5-year survival rates for all patients were 45.3 ± 5.1% and 41.8 ± 5.8%, and for those who survived the first 90 postoperative days 75.8 ± 6.1% and 70.0 ± 8.0%, respectively. The overall 5-year freedom from reoperation was 85.1 ± 5.7%. Preoperative predictors for 30-day mortality were Staphylococcus aureus [odds ratio (OR) 1.65; P = 0.04] and left ventricular ejection fraction (LVEF) <35% (OR 12.06; P = 0.03), for 90-day mortality acute kidney injury requiring dialysis (OR 6.2; P = 0.02) and LVEF <35% (OR 9.66; P = 0.03) and for long-term mortality cardiogenic shock (hazard ratio 2.46; P = 0.01).

CONCLUSIONS

Double valve replacement with reconstruction of the IFB in patients with complex IE is a challenging operation associated with high morbidity and mortality, particularly in the first 90 days after surgery. Survival and freedom from reoperation rates are acceptable thereafter, particularly considering the severity of disease and complex surgery.

Infective endocarditis, Mitral and aortic valve replacement, Intervalvular fibrous body

INTRODUCTION

One of the most dreaded complications of infective endocarditis (IE) is the extension of the infectious process into the periannular tissues resulting in the development of paravalvular abscesses with involvement and destruction of the intervalvular fibrous body (IFB). It requires complex surgical and intensive care management and is associated with high mortality rates.

The cornerstone of this procedure is the thorough debridement of all infected tissues and/or all prosthetic material combined with drainage of all abscess cavities, thus creating a large defect that leads to a lack of necessary anchoring support for a conventional heart valve operation. This large defect can only be corrected through complex reconstructive surgery, comprising mitral and aortic valve replacement with patch reconstruction of the IFB. Depending on the extent of involvement of the aortic root and its surrounding structures, root replacement may also be necessary. This challenging surgical procedure is associated with a high perioperative morbidity and mortality [1–3] which ranges between 20% and 30% in literature [1–5].

Whether the 65–70% hospital survival rate translates into similarly poor long-term outcomes is unknown, because late results following this procedure have not been reported. Therefore, the aim of this study was to assess the 3- and 5-year outcomes in patients undergoing double valve replacement (DVR: aortic and mitral valve replacement) with patch reconstruction of the IFB for treatment of IE extending into the IFB. Our hypothesis was that patients who survive the initial difficult perioperative phase would have an acceptable long-term event-free survival.

MATERIALS AND METHODS

Ethical statement

The study was approved by the ethics committee of the faculty of medicine at the University of Leipzig. Individual patient informed consent was waived.

Study design

Between January 1999 and December 2018, a total of 906 patients underwent combined aortic and mitral valve surgery for IE at our institution. Amongst them, 127 patients (14.0%) had concomitant involvement of the IFB and underwent radical surgical debridement of all infected tissue, followed by patch reconstruction of the IFB and DVR with bovine pericardium. These patients form the basis of the present study. Patients who underwent combined aortic and mitral valve surgery without IFB reconstruction or those who underwent aortic valve replacement and mitral valve repair with or without IFB reconstruction were excluded (Supplementary Material, Fig. S1). Primary outcomes were long-term survival and long-term freedom from reoperation. Secondary outcomes were 30- and 90-day mortality.

Surgical technique

The details of our technique have been described previously [5]. Briefly, the operation is performed through a median sternotomy and cardiopulmonary bypass. Standard cardiotomy incisions include an oblique aortotomy extending through the non-coronary aortic sinus and annulus, as well as a longitudinal incision in the left atrial roof extending towards the right superior pulmonary vein ostium. Following excision of the native/prosthetic aortic valve, an en bloc resection of the area of confluence of the aortic root, the left atrial roof and the base of the anterior mitral leaflet is performed, if possible. This en bloc specimen also includes the IFB. The extent of resection or the dimension of the resected tissue depends upon the macroscopic extent of the infection, and should preferably include a so-called ‘infection-free’ margin of 3–5 mm. Therefore, thorough excision and debridement of all infected valve tissues, all prosthetic valve material as well as drainage of all abscess cavities results in a common left ventricular orifice instead of inflow and outflow tracts (Fig. 1A).

Figure 1:

Surgical technique. (A) Single orifice of the LV. The yellow curved hatched line depicts the posterior mitral annulus between the LA and LV. (B) Mitral annular sutures (yellow hatched arrows) passed through the posterior mitral annulus and the mechanical prosthesis (black arrow). (C) Mitral prosthesis (white arrow) fixed to the posterior annulus from lateral to medial fibrous trigones (*). Margins of the left atrial roof shown with yellow arrows. (D) Anterior (white arrow) and posterior limbs (black arrow) of the folded pericardial patch that is sutured to the anterior sewing cuff of the mitral prosthesis. Posterior limb is sutured to the left atrial roof (yellow hatched arrows). (E) Anterior limb of the pericardial patch used to construct left ventricular outflow tract. Aortic annular sutures and mitral prosthesis seen in the background (yellow hatched arrow). Ao: aorta; LA: left atrium; LCA: left coronary artery; LV: left ventricle; RCA: right coronary artery; SVC: superior vena cava.

Thereafter, the new mitral mechanical or bioprosthesis is implanted, being initially secured only to the posterior mitral annulus from the lateral to the medial fibrous trigones (Fig. 1B and C). The anterior mitral annulus, which forms a margin of the IFB, is then reconstructed with a bovine pericardial patch, which is folded upon itself. The folded margin of the patch is sewn to the exposed (anterior third) sewing cuff of the mitral prosthesis. The posterior limb of the patch is then sewn to the free margins of the resected left atrial roof (Fig. 1D). Subsequently, the anterior limb of the patch is sutured to the margins of the left ventricular outflow tract and superiorly to the incised margins of the ascending aortic wall, thus forming the new IFB/subaortic curtain and the expanded non-coronary sinus, respectively (Fig. 1E). Thereafter, aortic valve replacement is performed. Extensive involvement of the aortic root and its surrounding structures by the infectious process necessitates drainage of all abscess cavities and radical debridement of all infected tissues and prosthetic material resulting in a large defect that makes additional aortic root replacement necessary. The aortic valve/root prostheses are anchored to the aortic annulus/left ventricular outflow tract, as well as to the anterior limb of the pericardial patch. In case of aortic valve replacement, the patch is used to close the right side of the aortotomy. When aortic root replacement with a homograft is used, a single pericardial patch is used to close the left atrial roof, and the IFB/subaortic curtain is reconstructed by sewing the anterior mitral leaflet of the homograft to the anterior third of the sewing cuff of the mitral prosthesis.

Data collection and follow-up

The demographic profile of patients, intraoperative data and postoperative outcomes were prospectively collected and entered into a computerized database and were retrospectively analysed. Follow-up was performed by mailed questionnaires or by phone contact with patients and/or close family members, with Supplementary Material supplied by family physicians and referring cardiologists. The closing interval for this study was between March and April 2019.

Statistical analysis

Categorical variables are expressed as frequencies and percentages throughout the manuscript. Continuous variables are expressed as mean ± standard deviation for normally distributed variables, and as median and interquartile range for non-normally distributed variables. Perioperative variables that had a univariable value of P < 0.25 or those judged to be clinically important were submitted to a multivariable logistic regression model by backward stepwise selection to determine the independent predictors of 30- and 90-day mortality, which are expressed as odds ratios (ORs) and 95% confidence intervals (95% CIs). Model calibration was performed by the Hosmer and Lemeshow good-of-fit test. Crude survival and freedom from reoperation were estimated with Kaplan–Meier survival methods. Survival estimates for the entire cohort are presented at 1- and 5-year intervals. A total of 10 patients who could have a follow-up >5 years were lost to follow-up and were censored in the Kaplan–Meier analysis at the point of the last follow-up date. In order to account for the impact of these patients lost to follow-up on the overall survival, a sensitivity analysis was performed, whereby 2 survival curves were compared; one assuming that they were alive until the time of the closing interval of the follow-up and the other assuming that the patients were dead at the last contact date. In order to determine the clinical factors associated with long-term mortality, a univariate Cox proportional hazards regression to evaluate the association of important clinical variables with mortality was performed. A multivariable Cox proportional hazard model using a combination of forward selection and clinical relevance was developed by initially selecting the same variables (used in univariate analysis) and entering them into the multivariable Cox regression model. Additional variables were then added and nested models were compared with the analysis of variance test to determine the incremental benefit of each additional included variable. When the analysis of variance P-value was >0.05, the manual stepwise selection was concluded. The results of the Cox model are reported as hazard ratios (HRs) with 95% CI. Proportional hazards assumption was statistically and graphically assessed using scaled Schoenfeld residuals against the transformed time (Supplementary Material, Table S1 and Fig. S2). All analyses were performed using SPSS Statistics (IBM Corp. Released 2017. IBM SPSS Statistics for Macintosh, Version 25.0. Armonk, NY, USA: IBM Corp.) and R 3.6.3 (The R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Demographic characteristics

Patient demographic characteristics are depicted in Table 1. The patients’ age ranged between 19 and 87 years. The main indications for surgery were preoperative detection of paravalvular abscesses and decompensated heart failure in 72 (56.7%) patients and 47 (37.0%) patients, respectively. In 15 (11.8%) patients, a paravalvular abscess was first detected intraoperatively, so that a total of 87 (68.5%) patients had confirmed paravalvular abscesses. Microorganisms could be isolated in 63.8% of the blood samples. They were Staphylococcus aureus in 24 (18.9%), Staphylococcus epidermidis in 8 (6.3%), other Staphylococcus species in 5 (3.9%), Enterococcus faecalis in 16 (12.6%), Streptococci in 18 (14.2%), Candida species (spp.) in 2 (1.6%), Propionibacterium acnes in 5 (3.9%) and Gram-negative bacteria in 3 (2.4%). A total of 46 (36.2%) patients had culture-negative endocarditis. The focus of infection was known in 49 (38.6%) patients only. Of these, 12 (9.4%) patients had pulmonary infections, 3 (2.4%) patients had gastrointestinal tract/intraabdominal infections, 13 (10.2%) patients had infected skin wounds, 5 (3.9%) patients had urinary tract infections, 4 (3.1%) patients had ear, nose and throat infections, 4 (3.1%) patients had spondylitis, 7 (5.5%) patients had indwelling catheters and 1 (0.8%) patient was an intravenous drug abuser. Septic embolization occurred in 52 (40.9%) patients, of which 20 (15.7%) patients were multiorgan (≥2 body organs), 17 (13.4%) patients had cerebral and 15 (11.8%) patients had splenic embolisms.

Table 1:

Demographic characteristics

Preoperative variables
Age (years)	65.3 ± 12.9
Male gender	98 (77.2)
Body mass index (kg/m2)	27.8 ± 5.4
Systemic arterial hypertension	112 (88.2)
Diabetes mellitus	49 (38.6)
Dyslipidaemia	85 (66.9)
Chronic obstructive pulmonary disease	12 (9.4)
Acute kidney injury requiring dialysis	13 (10.2)
Peripheral vascular disease	22 (17.3)
Recent stroke caused by septic embolism	27 (21.3)
Sepsis	24 (18.9)
Septic embolism	52 (40.9)
NYHA class
I	10 (7.9)
II	26 (20.5)
III	66 (51.9)
IV	25 (19.7)
Left ventricular ejection fraction (%)	53.7 ± 10.3
>50%	72 (56.7)
36–50%	47 (37.0)
≤35%	8 (6.3)
Previous pacemaker implantation	27 (21.3)
Cardiogenic shock	22 (17.3)
Mechanical ventilation	4 (3.1)
Coronary artery disease	48 (37.8)
1-vessel	23 (18.1)
2-vessel	10 (7.9)
3-vessel	15 (11.8)
Native valve endocarditis	29 (22.8)
Aortic valve	6 (4.7)
Mitral valve	4 (3.1)
Aortic and mitral valves	19 (14.9)
Prosthetic valve endocarditis	98 (77.2)
Early prosthetic valve endocarditis (<1 year)	35 (27.6)
Late prosthetic valve endocarditis (>1 year)	63 (49.6)
Aortic valve	41 (32.3)
Mitral valve	2 (1.6)
Aortic and mitral valve	55 (43.3)
Previous cardiac surgery	104 (81.8)
Aortic root replacement	16 (12.6)
Aortic valve replacement	49 (38.6)
Mitral valve replacement	5 (3.9)
Aortic and mitral valve replacement	15 (11.8)
Mitral valve repair	2 (1.6)
Aortic valve replacement + CABG	13 (10.2)
ASD closure	2 (1.6)
Isolated CABG	2 (1.6)
Timing of surgery
Urgent	49 (38.6)
Emergent	78 (61.4)
Logistic EuroSCORE (%)	53.0 ± 25.2

Continuous variables expressed as mean ± standard deviation. Categorical variables expressed in numbers (n) and percentages in parentheses.

ASD: atrial septal defect; CABG: coronary artery bypass grafting; NYHA: New York Heart Association.

Intraoperative data and postoperative outcomes

Intraoperative details have been depicted in Table 2. Tricuspid valve (TV) repair or replacement was performed in 31 patients. Of these, 5 (3.9%) patients had findings suggestive of TV endocarditis and 8 patients had pre-existing tricuspid regurgitation due to annular dilatation 5 (3.9%) or pacemaker wires 3 (2.3%). The remaining 18 patients (14.2%) required TV surgery as a corrective measure for injury/deformation of the TV annulus that occurred during surgery.

Table 2:

Intraoperative data

Intraoperative details
Aortic valve replacement	127 (100)
Mechanical	26 (20.5)
Stented biological	85 (66.9)
Stentless biological	10 (7.8)
Homograft	6 (4.7)
Aortic root replacement	75 (59.1)
Mitral valve replacement	127 (100)
Mechanical	26 (20.5)
Biological	101 (79.5)
Concomitant procedures
Atrial septal defect closure	3 (2.3)
Tricuspid valve replacement	4 (3.1)
Tricuspid valve repair	27 (21.3)
Partial arch replacement	4 (3.1)
Coronary artery bypass graft surgery	28 (22.0)
Cardiopulmonary bypass time (min)	220.8 ± 73.8
Aortic cross-clamp time (min)	162.4 ± 41.3
Paravalvular abscessesa	87 (68.5)

Continuous variables expressed as mean ± standard deviation. Categorical variables expressed in numbers and percentages in parentheses.

a

Abscesses confirmed intraoperatively.

A total of 36 patients (28.3%) died within 30 days following surgery. Of these, 8 (6.3%) patients died in the operating room; 4 (3.1%) each due to irreversible circulatory failure and uncontrollable bleeding. Other deaths occurred because of cardiogenic shock (n = 11; 8.7%), multiorgan system failure (n = 8; 6.3%), septic shock (n = 5; 3.9%) and other non-cardiac causes (n = 4; 3.1%). Multivariable logistic regression analysis identified S. aureus positive blood cultures (OR 1.65; P = 0.04) and left ventricular ejection fraction (LVEF) <35% (OR 12.06; P = 0.03) as independent predictors for 30-day mortality and acute kidney injury requiring dialysis (OR 6.2; P = 0.02) and LVEF <35% (OR 9.66; P = 0.03) for 90-day mortality. Native valve IE was found to be protective (OR 0.005; P = 0.001 for 30-day mortality and OR 0.12; P = 0.001 for 90-day mortality). The postoperative outcomes are summarized in Table 3. Small aortic and mitral paravalvular leaks were observed in 6 and 7 patients, respectively, on predischarge echocardiography. They were neither haemodynamically relevant nor did they produce any haemolysis and were therefore conservatively managed.

Table 3:

Postoperative outcomes

Outcomes
30-day mortality	36 (28.3)
90-day mortality	47 (37.0)
Low cardiac output	23 (18.1)
Extracorporeal membrane oxygenation	19 (14.9)
Sepsis	26 (20.5)
Re-exploration for bleeding	25 (19.7)
Atrial fibrillation	78 (61.4)
Pacemaker implantation	53 (41.7)
Stroke	11 (8.7)
New dialysis	55 (43.3)
Respiratory failure requiring reintubation	13 (10.2)
Tracheostomy	30 (23.6)
Hospital stay (days), median (IQR)a	19 (8.5–33.0)

Categorical variables expressed in numbers and percentages in parentheses.

a

Reported as median and interquartile range (IQR) in parentheses.

Follow-up

A total of 11 patients died between 30 and 90 days following surgery, resulting in a 90-day mortality of 37.0%. These deaths occurred due to multiorgan failure secondary to previous septic shock (n = 9; 7.1%), acute respiratory distress syndrome (n = 1; 0.8%) and other non-cardiac causes (n = 1; 0.8%). The mean follow-up was 477.5 ± 799.2 months and ranged from 0 to 136.4 months. A total of 13 (10.2%) deaths occurred during follow-up after the first 90 postoperative days. Among them, 3 (2.4%) patients had cardiac deaths, 4 (3.1%) patients died due to multiorgan system failure, 1 (0.8) patient died due to sepsis following recurrent endocarditis and 5 (3.9%) patients died due to other non-cardiac causes. The estimated overall 3- and 5-year survival rates were 45.3 ± 5.1% and 41.8 ± 5.8%, respectively (Table 4). The corresponding survival rates in patients who survived the first 90 postoperative days were 75.8 ± 6.1% and 70.0 ± 8.0% (Fig. 2), which demonstrates that the perioperative period has the highest hazard for mortality, which declines through the first year and stabilizes thereafter (Supplementary Material, Fig. S3). The vast majority of patients included in the present study underwent surgery in the last 5–6 years. Of the 28 patients who could have a follow-up longer than 5 years (patients operated between 1998 and 2012), 10 were lost to follow-up. Sensitivity analysis that compared 2 survival curves developed on the assumption that these 10 patients were alive vs dead demonstrated no significant difference (Supplementary Material, Fig. S4). A total of 9 (7.0%) patients required reoperations, 7 due to recurrent IE and 2 due to other causes. All but 2 reoperations were re-reconstruction of the IFB with DVR. Estimated 5-year freedom from reoperation was 85.1 ± 5.7% (Fig. 3). Cox regression analysis identified cardiogenic shock (HR 2.46; P = 0.01) as the only preoperative factor associated with long-term mortality. Compared to patients with prosthetic valve endocarditis, patients with native valve IE (HR 0.25; P = 0.004) had significantly better survival as also did those who had suffered a recent preoperative stroke caused by septic embolism (HR 0.37; P = 0.01) (Table 5).

Figure 2:

Kaplan–Meier curves depicting long-term survival in all patients (lower curve) and in patients who survived the first 90 postoperative days (upper curve).

Figure 3:

Kaplan–Meier curve depicting long-term freedom from reoperation in all patients.

Table 4:

Follow-up results

Long-term outcomes
Estimated 3-year survival (%)	45.3 ± 5.1
Estimated 5-year survival (%)	41.8 ± 5.8
Recurrent endocarditis, n (%)	7 (5.5)
Reoperations due to recurrent endocarditis, n (%)	7 (5.5)
Reoperations due to other causes, n (%)	2 (1.6)
Estimated 5-year freedom from reoperation (%)	85.1 ± 5.7

Table 5:

Predictors of late mortality

Preoperative parameters	Hazard ratio	95% CI	P-value
Age	1.06	0.99–1.04	0.16
Cardiogenic shock	2.46	1.21–4.99	0.01
Native valve endocarditis	0.25	0.1–0.65	0.004
Recent stroke caused by septic embolism	0.37	0.16–0.82	0.01
Female gender	0.64	0.32–1.28	0.21
Acute kidney injury requiring dialysis	1.26	0.58–2.73	0.55
Culture-negative endocarditis	1.13	0.67–1.93	0.62

All variables were included in a single regression model.

CI: confidence interval.

DISCUSSION

Optimal management of an IFB destroyed by IE is still controversial [1, 5–7] and represents only a minority of patients (14.0%) undergoing surgery for IE at our institution. The Toronto group, which was one of the pioneers of the surgical technique used to manage such pathology [1], has always emphasized the importance of radical eradication of infected tissue and considers it to be the cornerstone in achieving satisfactory long-term results [3]. Corrective surgery, being technically challenging, has a prohibitively high early mortality, but remains the only option available for this severely ill group of patients, who otherwise face certain death. Such procedures are associated with a high mortality, even when only one valve is affected [8]. Our 30- and 90-day mortality (28.3%, 37.0%) for such procedures is comparable to that reported by the Toronto group (20–30%) [4, 8]. We have previously demonstrated that additional mitral valve endocarditis in patients with aortic root abscesses is associated with a higher mortality in comparison to those who do not (36.2 vs 20%, P = 0.04) [9]. Contiguous spread of infection from the aortic to the mitral valve and double valve surgery have been found to be predictors of early death [10].

Our estimated 5-year survival rate was 41.8% but was chiefly driven by the high early mortality. Most deaths occurred within the first 3 months following surgery, after which the survival rates improved. Such patients are at a high risk of perioperative refractory sepsis and vasoplegia because of intraoperative contamination and protracted cardiopulmonary bypass and operative times or later in the intensive care unit due to persistent bacteraemia or superinfections associated with prolonged ventilation and requirement of central venous and/or dialysis catheters. Recently, the Cleveland clinic group reported an in-hospital mortality of 24% and survival at 1, 5, and 7 years of 60%, 40% and 35%, respectively, and concluded that surgery provides the only chance for cure [11]. However, patients in whom the mitral valve was preserved showed an insignificant trend towards a lower in-hospital mortality (13%) and better survival (80%, 64% and 57%). The authors attributed these observations to preservation of the mitral valve and the subvalvular apparatus leading to maintenance of left ventricular function [11, 12]. Better outcomes could also be due to performance of less complex surgery in patients having less severe disease (lower incidence of vegetations, mitral annular abscesses or intracardiac fistulae). Therefore, tailoring the operation to the pathology and the patient is particularly important in securing better early and late outcomes.

Our study importantly shows that overall long-term survival is directly dependent upon perioperative mortality, which is predominantly associated with the degree of sepsis and tissue destruction that are frequently severe following S. aureus infections and often result in complications such as acute kidney injury, septic cardiomyopathy and left ventricular dysfunction, all predictors of 30- and 90-day mortality. However, following successful treatment of sepsis and its detrimental effects, survival significantly improved in patients who survived the first 3 months (Fig. 2 and Supplementary Material, Fig. S3). This could also explain why cardiogenic shock, which is usually observed in patients with poor left ventricular function and severe acute valvular regurgitation following IE, was overridden by factors related to sepsis as a predictor of death during the early perioperative phase but emerged as the only factor associated with long-term death. Therefore, an ‘Endocarditis team’, which can address the vast majority of comorbidities and complications associated with the disease and complex surgery, is extremely essential to achieve satisfactory outcomes [13]. It is chiefly responsible in determining the indication and timing of surgery based on the patient’s preoperative status, comorbidities, clinical features, patient’s life expectancy, operative risk and likelihood of postoperative recovery. Preoperative planning with the help of imaging modalities such as transoesophageal echocardiography and thoracic computed tomographic scans are imperative to formulate an effective treatment strategy [14]. The value of positron emission tomography in diagnosing subtle cases of IE cannot be overemphasized. Haemodynamic monitoring and stability, surgical dexterity and extensive experience of the operating room team including surgeons, anaesthesiologists, perfusionists and nurses are essential for achieving optimal outcomes. The surgeon’s ability to recognize and excise all infected and devitalized tissues and effectively reconstruct the resulting defect cannot be overstated. Furthermore, excellent postoperative multidisciplinary management in the intensive care unit is of utmost importance.

Referral of patients for surgical management often ensues only after medical management has failed, leading to deterioration of the patients’ general clinical condition, as evidenced by the strikingly high mean EuroSCORE of our patients. The emergence of recent stroke caused by septic embolism as a protective factor for late mortality could probably be due to the fact that such patients were diagnosed and operated earlier than other patients, resulting in better outcomes. Therefore, early surgery should be advocated as results have been found to be superior, particularly in patients with infections caused by S. aureus and other highly virulent microorganisms that can penetrate and destroy perivalvular tissues very rapidly [15, 16]. Staphylococcus species have been previously identified as an independent predictor of poor outcomes [2]. This correlates well with our study population, in which 30 out of 37 (81.1%) patients with Staphylococcal infections presented with a paravalvular abscess. Furthermore, S. aureus was an independent predictor of 30-day mortality, probably because almost half of the patients infected with Staphylococcus spp. died.

High prevalence of patients who have undergone previous cardiac surgery is also one factor that contributes to the high complexity and morbidity associated with this procedure. Native valve endocarditis was found to be a protective factor for early and late mortality, which is a surrogate marker for either prosthetic valve endocarditis or reoperation being potent risk factors. The high mortality rate following reoperations (55.3%) could not only be due to the greater technical difficulty posed by severe adhesions and complex pathology but also because such patients usually require reoperations early following their previous surgery (<1 year) and are often extremely sick due to early development of abscesses, multiple comorbidities and a compromised general clinical status, not having yet fully recovered from their previous operations.

The impact of the type of prostheses used as replacements on long-term outcomes in patients with IE is still a topic of debate. Homografts have frequently been the preferred option for patients with paravalvular abscesses because of a lower incidence of recurrent endocarditis, its resistance to infection and better penetration of antibiotics [17, 18], good biomechanical performance with low transvalvular gradients [19], better haemostasis during surgery and the possibility of using the anterior mitral leaflet of the aortic homograft for reconstruction of the aorto-mitral continuity [11, 12]. However, they are prone to degeneration due to progressive calcification [20] and reoperations are difficult due to the development of dense adhesions. Additionally, several reports in literature have found in-hospital mortality, survival and the long-term freedom from reinfection or reoperation due to recurrent IE to be independent of valve type [21–23]. The Cleveland group recently reported their 5-year freedom from reoperation and recurrent endocarditis rates of 63% and 77%, respectively, with 45% of patients receiving allografts [11]. Our study revealed an 85% freedom from reoperation at 5 years, with 4.7% of patients receiving homografts. Therefore, the valve choice in patients requiring IFB reconstruction should theoretically be the same as for any patient requiring valve replacement for non-infectious causes and should be individualized for every patient. Radical resection rather than valve type is the key to prevention of recurrent/residual infection.

Limitations

This study, being a single-centre experience and retrospective in nature, is subject to inherent biases. The study cohort is small; thus, interpretation of various statistical analyses should be addressed with caution. However, for a rare disease that requires complex surgery, our experience would be the largest in literature to date. Being a tertiary referral centre that performs large volumes of surgeries in patients with IE, our current results may not be applicable to all institutions, especially those that have very limited experience in managing IE. Finally, we do not have data regarding complications such as paravalvular leaks, pseudoaneurysms, etc. due to the lack of echocardiographic follow-up.

CONCLUSION

DVR with reconstruction of the IFB in patients with complex IE is a challenging operation associated with high morbidity and mortality. However, associated outcomes are acceptable as it is the only available option for patients who otherwise face 100% mortality. The highest probability of death exists within the first 90 days following surgery, after which survival and freedom from reoperation are better, considering the severity of the disease and surgical complexity.

Supplementary material is available at EJCTS online.

Conflict of interest: none declared.

Author contributions

Piroze M. Davierwala: Conceptualization; Methodology; Project administration; Supervision; Validation; Visualization; Writing—review & editing. Mateo Marin-Cuartas: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Software; Supervision; Validation; Visualization; Writing—original draft; Writing—review & editing; Oral presentation EACTS Lisbon 2019. Martin Misfeld: Conceptualization; Data curation; Investigation. Salil V. Deo: Formal analysis; Methodology; Supervision; Statistical review. Sven Lehmann: Conceptualization; Data curation; Investigation. Jens Garbade: Conceptualization; Data curation; Investigation. David M. Holzhey: Conceptualization; Data curation; Investigation. Michael A. Borger: Conceptualization; Formal analysis; Investigation; Methodology; Supervision; Validation; Visualization; Writing—review & editing. Farhad Bakhtiary: Supervision; Validation; Visualization; Writing—review & editing.

Presented at the 33rd Annual Meeting of the European Association for Cardio-Thoracic Surgery, Lisbon, Portugal, 3–5 October 2019.

ABBREVIATIONS

 
• CI

  Confidence interval

 
• DVR

  Double valve replacement

 
• HR

  Hazard ratio

 
• IE

  Infective endocarditis

 
• IFB

  Intervalvular fibrous body

 
• LVEF

  Left ventricular ejection fraction

 
• OR

  Odds ratio

 
• TV

  Tricuspid valve

Author notes

Piroze M. Davierwala and Mateo Marin-Cuartas contributed equally to this work.