Prosthetic valve endocarditis: predictors of early outcome of surgical therapy. A multicentric study

Abstract

INTRODUCTION

Prosthetic valve endocarditis (PVE) is probably the most serious complication in patients with prosthetic valve. This problem was pointed out over half a century ago when cardiac valve replacements were spreading worldwide [1].

PVE occurs in 1–6% of patients with valve prosthesis and accounts for 16–34% of all cases of infective endocarditis [2–4]. Both tissue and mechanical valves could be equally affected [5–7].

This life-threatening event is associated with significant mortality ranging from 20% to 80% despite improvements in diagnosis, antimicrobial drugs and surgical technique. According to the guidelines [8], surgery is the treatment of choice in complicated PVE depending on clinical course, prosthesis dysfunction, periannular involvement and aetiology. Nevertheless the global therapeutic strategy such as the role of conservative treatment and optimal timing of operation is still a matter of debate [9].

The aim of our study was to investigate the aetiology, to identify clinical features and to determine the risk factors for adverse early outcome after surgery in PVE patients receiving surgical treatment over a 14-year period.

METHODS

Study design

A retrospective cohort study was conducted by enrolling patients aged ≥18 diagnosed with PVE in accordance to modified Duke criteria. Based on ESC guidelines, PVE occurring within 1 year from primary valvular surgery was classified as ‘early’ while after 1 year as ‘late’ [8].

All data concerning patient and infection characteristics, preoperative comorbidities and postoperative events were obtained from the medical records of case patients.

The primary outcome measured was in-hospital (death prior to patient discharge at any time) and intraoperative (death when occurred in the operating theatre) mortality. In-hospital mortality engulfs intraoperative and 30-day mortalities.

Postoperative morbidity, recurrences and need for redo-operation were assessed as secondary outcomes. Survivor and non-survivor subgroups were compared to identify predictors of in-hospital mortality.

Study approval was granted by the Research Ethics Committee of the coordinating centre (Catholic University of the Sacred Heart) and it was further communicated and extended to the participating centres. Informed consent was waived due to the retrospective observational nature of the study.

Study setting

Data were obtained from PVE cases consecutively admitted to 8 Italian healthcare facilities from 1 January 2000 to 31 December 2013.

Definitions

The following terms were defined prior to data collection. Hospital admission was defined as the date of patient’s admission to the study facility. Patient comorbidities and postoperative complications were defined according to the joint AATS/EACTS/STS adult cardiac surgery committee [10]. Embolic events were defined as clinical or radiographic evidence of cerebral or peripheral involvement. Stroke was defined as a prolonged focal/global neurologic deficit lasting ≥72 h. Critical conditions were defined in the presence of at least one of the following: persistent sepsis despite antimicrobial treatment, new onset of conduction abnormality, severe prosthetic dysfunction paired with haemodynamic instability (New York Heart Association functional Class III and IV heart failure)

Shock was defined as persistent hypotension ≤90 mmHg associated at least with one of the following: oliguria, acute alteration of mental status or lactic acidosis. Echographic findings of valvular and perivalvular complications were defined as the presence of any vegetation, abscess or fistula by transthoracic and/or transoesophageal echocardiography. Renal impairment indicated creatinine level closest to the date and time prior to surgery and was defined as serum creatinine ≥2 mg/dl. Urgent surgery was defined as surgery performed within the first week from admission. Elective surgery was instead considered when performed later than 1 week from admission [11]. The infecting organism was determined by blood cultures and/or valvular cultures at the time of diagnosis or surgery. Cases were defined as culture-negative PVE if no microorganism could be identified [12]. Major surgical operation indicated previous major surgical procedures (abdominal, orthopedic, neurological and gynaecological, etc) performed before PVE occurrence.

Statistical analysis

Results were expressed as means ± standard deviation or medians (interquartile range) (continuous variables) or as percentages of the group from which they were derived (categorical variables). The Student t-test and Mann–Whitney U-test were used to compare normally and non-normally distributed continuous variables, respectively. Differences in group proportions were assessed by using the χ² test or Fisher’s exact test in case of small numbers. Odds ratio and 95% confidence intervals were calculated for all emerged associations. Multivariable logistic regression analysis was used to identify independent risk factors for in hospital mortality. Variables emerging from univariable analysis with P-values of <0.1 were included in the multivariable model in a backward stepwise manner. All statistical analyses were performed with Stata11 (StataCorp LP, College Station, TX, USA).

RESULTS

During the study period among 46 361 patients undergoing isolated or combined valve procedures, PVE was the surgical indication in 209 consecutive patients (0.45%). This rate significantly increased during the second time frame (2007–2013) compared to the first one (2000–2006) (0.58% vs 0.31%; P < 0.001). The median patients’ age was 68 (interquartile range 56–74), 141 (67.5%) were male, New York Heart Association functional Class III and IV heart failure was reported in 144 patients (68.9%).

A total of 55 patients (26.3%) sustained 1 or multiple preoperative embolic complications involving 1 or more systems: brain including retina, mesenteric circulation including spleen, upper and lower limbs, kidney and coronary arteries.

Preoperative clinical status was characterized by shock in 25 cases (11.2%), renal impairment in 33 cases (15.8%) and pharmacological circulatory support in 37 cases (17.7%).

The onset of PVE was early in 89 patients (42.6%) and late in 120 (57.4%). The median duration from diagnosis of PVE to operation was 16 days (range 7–35).

Infection involved mechanical valves in 114 cases (54.5%) and bioprosthetic ones in 98 cases (48.9%). The aortic valve was affected in 139 patients (66.5%), the mitral in 85 (40.7%) and the tricuspid in 6 patients (2.9%). Fifteen patients (7.2%) had multivalvular involvement. Other preoperative demographics, comorbidities, baseline clinical characteristics, echocardiographic and electrocardiographic findings of cohort patients are listed in Table 1.

A pathogenic micro-organism was isolated in 180 (86.1%) patients. Table 2 shows the detailed microbiological results.

Antibiotic therapy based on guidelines directed by microbiological findings or empirical in case of culture-negative PVE was administered for a total of 4–6 weeks according to current recommendations [8]; the timing was established at the discretion of the treating physician at each Institution.

The indications for operation in PVE patients were: paravalvular leak in 124 (59.3%) cases, abscess in 68 (32.5%), prosthetic valve dysfunction leading to cardiac failure in 42 (10 tissue valve perforation and 32 stenosis), large mobile vegetation in 20 (9.5%), systemic embolization in 55 (26.3%) and persistent fever in 54 (25.8%). Many patients had more than 1 indication for reoperation. The median interval between the previous cardiac operation and the redo procedure was 15 months (range 1–118 months).

Reoperation for PVE was the first redo in 181 (86.6%) patients, the second redo in 21 (10%) and the third redo in 7 (3.3%) considering the clinical history of each patients.

The operative data, postoperative complications, median time of intensive care unit and hospital stays (in total and according to in-hospital and intraoperative mortality) are listed in detail in Table 3.

Intraoperative, 30-day and in-hospital mortality rates of all 209 patients were as follows: 9 (4.3%), 42 (20.6%) and 45 (21.5%).

A comparison between the survivor and non-survivor subgroups at univariable analysis is shown in Table 4. In addition, among non-survivor features, prior invasive procedures were found to be significantly higher in the late PVE cohort (3/18 vs 0/27 P = 0.02). On the contrary, paravalvular leaks were more common in patients with early PVE (24/27 vs 1/18 P < 0.001).

Multivariable logistic regression analysis identified female sex, shock status, major previous surgical procedures within the last 3 months, multivalvular intervention, periprosthetic abscess and urgent indication as independent risk factors for mortality in PVE patients who underwent surgical intervention (Table 5). The model had an area under the ROC curve of 0.86, indicating good predictive ability and absence of multicollinearity.

Follow-up (overall survival, freedom from reoperation and re-endocarditis) was gathered by direct contact with patients, their physicians or both and obtained in 157 (95.7%) out of 164 surviving patients. Median follow-up was 52 ± 9 months. At 36 months follow-up, rates of freedom from reoperation and from endocarditis were 152/164 (92.6%) and 155/164 (94.5%), respectively. Overall survival of all 209 patients at 12 and at 36 months was 160/209 (76.5%) and 142/209 (67.9%), respectively. Out of 200 patients surviving the intervention, overall survival for the mentioned time points was 80% and 71%, respectively.

DISCUSSION

PVE is a life threatening complication in patients with prosthetic valve [9, 13, 14]. Edwards et al. [13] reported an incidence ranging from 0.3% to 0.8% and these values remained stable during the study period (1986–1996). In our surgical experience, among the valve diseased population, the rate of PVE cases during the study period increased significantly: 0.31% in the first half to 0.58% in the second half (P < 0.01).

This significant increase could be explained by the high number of vulnerable patients carrying prosthetic valves who frequently require invasive diagnostic and therapeutic procedures (3.8% in haemodialysis, 26.8% carrying intravascular devices) in light of the prolonged life expectancy which has occurred over the last decades.

In 52% of cases, the only predisposing factor for infection was the presence of prosthetic valve itself and no difference was found between mechanical versus biological devices [9, 14] having similar susceptibility to infection as reported by Baumgartner et al. [15].

The spectrum of micro-organisms responsible for PVE in our study was similar to other reported series, with Staphylococci accounting for more than 50% of cases [5, 9, 14, 16–19].

Surgical management is challenging and the main objectives are removal of infected necrotic tissue and foreign material, repair of cardiac anatomy and replacement of the prosthesis. In case of abscess, aggressive debridement is the cornerstone of treatment. Cryopreserved aortic homograft, employed in 15 patients of our series (7.2%), is a safe and effective solution in aortic PVE with periannular complications as shown by Sabik et al. in 103 patients with an impressive 3.9% in-hospital mortality. Other conventional prosthesis such as mechanical and biological ones, respectively representing 43.1% and 50.2% of our adopted solutions, can be used in the treatment of PVE. Furthermore, hospital mortality is not influenced by the type of valve implanted [20, 21].

In the early 1980s–90s series, surgical mortality for PVE ranged from 20% to 60%, [2] while 2 recent studies have reported a 30-day mortality below 15% [17, 19]. Table 6 shows that in the last decade hospital mortality after surgery for PVE remains significantly high: reaching 20% to 30% particularly in patients in critical clinical conditions or in complicated PVE [7, 16].

In our series, in-hospital mortality (21.5%) is comparable to other series and is predicted in multivariable statistical analysis by female sex, preoperative shock status, previous major surgical procedure, multivalvular involvement, abscess and urgent operation.

Multiple valve operation was the most significant determinant of mortality being a marker of a larger amount of an anatomic lesion with a longer procedure in terms of cardiopulmonary bypass and cross-clamp times.

In our population preoperative septic and/or cardiogenic shock is an independent predictor for poor early outcome. This finding is in accordance with many other surgical series [9, 18]. Heart failure [3, 5, 9, 14, 18], catecholamines infusion [5, 16], New York Heart Association IV functional class [3], severe reduction of left ventricular ejection fraction [3] and mechanical circulatory support [16] were important prognostic signs in PVE patients. All these variables and urgent operations led to an increasing mortality.

Abscess had a prevalence of 32.5% in our cohort and was a strong predictor of in-hospital mortality.

In several previous studies, periannular complications were associated with a significantly higher in-hospital mortality [3, 14, 16, 18, 22]. Accordingly, Anguera et al. [23] reported a 27% mortality rate in their large cohort of surgically treated PVE patients with periannular abscess.

Any major previous surgical procedure, carried out less than 3 months before diagnosis of PVE, resulted to be associated with in-hospital mortality. This association had not been identified before; we can speculate that a previous major surgical operation represents a potential source of infection and also decreases immunological defence. However we did not find any association with outcome and causative micro-organisms of PVE.

Concerning postoperative complications, re-exploration for bleeding occurred in 20% of our cases. This high rate is not surprising and is related to several factors: redo procedures, complex and long operation and inflammatory condition which can negatively affect coagulation cascade and interfere with a proper coagulative function.

Mid- and long-term survival rates greatly differ among studies: from 73% to 37% at 5 years [20, 24] and 75% to 31% at 10 years [5, 24]. This wide variability could be explained by a series of factors such as patient features or study newness. Our data are comparable with previous findings of other major case series; freedom from infection recurrence is satisfyingly high and in line with other reports.

Limitations

This study has some limitations: firstly, the retrospective observational nature; secondly, valvular redo procedures were performed by different surgeons in 8 cardiac surgery units with confounding results due to possible heterogenous surgical techniques adopted. Finally, our follow-up was complete in 95.7% of our cohort.

CONCLUSIONS

Critical preoperative conditions, extensive anatomic lesions and multivalvular involvement strongly contribute to increase in-hospital mortality which is still high ranging over 20% as shown by the most recent series and confirmed by us as well. Nevertheless in our experience, surgery provides excellent results to avoid reinfection or reoperation at both 12 and 36 months follow-up. A better understanding of preoperative variables associated with mortality may be helpful in the multidisciplinary decision-making process and family counselling in the event of PVE.

Conflict of interest: none declared.

REFERENCES