Infective endocarditis and spondylodiscitis—impact of sequence of surgical therapy on survival and recurrence rate

Abstract

OBJECTIVES

To date, there are no standardized treatment algorithms or recommendations for patients with infective endocarditis (IE) and concomitant spondylodiscitis (SD). Therefore, our aim was to analyse whether the sequence of surgical treatment of IE and SD has an impact on postoperative outcome and to identify risk factors for survival and postoperative recurrence.

METHODS

Patients with IE underwent surgery in 4 German university hospitals between 1994 and 2022. Univariable and multivariable analyses were performed to identify possible predictors of 30-day/1-year mortality and recurrence of IE and/or SD.

RESULTS

From the total IE cohort (n = 3991), 150 patients (4.4%) had concomitant SD. Primary surgery for IE was performed in 76.6%, and primary surgery for SD in 23.3%. The median age was 70.0 (64.0–75.6) years and patients were mostly male (79.5%). The most common pathogens detected were enterococci and Staphylococcus aureus followed by streptococci, and coagulase-negative Staphylococci. If SD was operated on first, 30-day mortality was significantly higher than if IE was operated on 1st (25.7% vs 11.4%; P = 0.037) and we observed a tendency for a higher 1-year mortality. If IE was treated 1st, we observed a higher recurrence rate within 1 year (12.2% vs 0%; P = 0.023). Multivariable analysis showed that primary surgery for SD was an independent predictor of 30-day mortality.

CONCLUSIONS

Primary surgical treatment for SD was an independent risk factor for 30-day mortality. When IE was treated surgically 1st, the recurrence rate of IE and/or SD was higher.

Open in new tabDownload slide

INTRODUCTION

Both infective endocarditis (IE) and spondylodiscitis (SD) are serious diseases associated with a high mortality rate of up to 30% [1–3]. Due to the haematogenous spread of infectious agents, simultaneous occurrence of IE and SD may occur [4].

The simultaneous occurrence of IE-SD coinfection has been little described and analysed [1, 4–8]. The incidence of SD in IE cohorts has also not been adequately studied, and reported rates vary significantly from 8% to 19% [4–7]. For SD cohorts, it has been previously reported that concomitant IE is present in 3% to 30% of SD cases [1, 9, 10]. The wide range of IE-SD coinfection rates may be attributed to differences in the routine performance rates of echocardiography in SD patients and spine magnetic resonance imagings in IE patients, respectively, depending on the primary inpatient department in which patients are admitted.

To date, there are no standardized treatment algorithms or recommendations for patients with IE and concomitant SD. Furthermore, there is little evidence on whether the concomitant presence of SD has a negative prognostic impact on the clinical course of IE and whether concomitant SD is a risk factor for postoperative survival and recurrence. In contrast, time to diagnosis has been shown to have a significant impact on prognosis and mortality in both IE and SD [1, 2, 11, 12].

To our knowledge, no study has investigated the impact of the sequence of surgical care (treat IE surgically 1st, then SD or vice versa). Therefore, our aim was to compare whether the sequence of surgical treatment of IE and SD has an impact on postoperative outcome, and to identify risk factors for survival and postoperative recurrence of IE and/or SD.

PATIENTS AND METHODS

Ethical statement

The study protocol was approved by institutional ethics committees (Ethics Committee of the Medical Faculty, University of Cologne, 17-407).

Patient selection, data collection and definition of variables

For this national multicentre retrospective analysis, data of 3991 consecutive IE patients who underwent surgery for IE in 4 German cardiac surgery centres between 1994 and 2022 were analysed. From the IE cohort, patients >18 years with concomitant SD were included in further analyses. The diagnosis of SD was discussed and confirmed by infectious disease specialists and orthopaedic surgeons on an interdisciplinary basis. Questionable cases were excluded. Data collected included patient demographics and concomitant diseases, manifestation of IE according to the recently modified Duke criteria [13] (echocardiographic and microbiological data), localization of affected spinal segment(s) based on conventional radiograph at initial admission, concomitant spinal manifestation (empyema, psoas abscess), perioperative data and relevant postoperative outcomes. IE-related complications were recorded for the hospital stay. Follow-up was determined by reviewing medical records and interviewing the patient or patient's physician. Recurrence was defined as the recurrence of IE and/or SD within the 1st year after surgery. Treatment failure was defined as death and/or recurrence within 1 year.

Chronic kidney disease was defined and classified according to the KDIGO classification (Kidney Disease—Improving Global Outcomes) and KDIGO 2012 guidelines [14]. Diabetes was diagnosed according to the American Diabetes Association criteria if the patient had a fasting glucose ≥126 mg/dl or a glycated haemoglobin (HbA1c) ≥6.5% if taking hypoglycaemic drugs [15]. Chronic obstructive pulmonary disease (COPD) is defined as a chronic and usually progressive airway/lung disease characterized by airway obstruction that is not completely reversible.

Surgical procedure

The indication for cardiac surgery for IE was made according to the current 2015 European Society of Cardiology (ESC) guidelines for the management of infective endocarditis [16]. The type of surgery was based on surgeon preference and the extent of infection. Intraoperative transoesophageal echocardiography was performed in all patients. Management of haemodynamics, catecholamines, and blood transfusions was at the discretion of the attending cardiac surgeon and anaesthesiologist.

The decision to surgically treat SD was based on well-defined criteria. Unless there was bony destruction, spinal instability, or empyema, a conservative therapeutic approach was chosen. The main reasons for surgical intervention were neurological deficits. However, progressive spinal deformities, (painful) spinal instability, intraspinal empyema, or unsuccessful conservative treatments also prompted surgery. All patients with progressive neurological deficits were immediately referred for surgical treatment.

Statistical analysis

Unless otherwise indicated, continuous variables are expressed as mean (± standard deviation) or median (interquartile range) according to the normal distribution. Normal distribution was assessed using the Shapiro–Wilk test. The t-test was used to compare the continuous variables if a normal distribution was present. If the requirements for a normal distribution were not met, the Mann–Whitney U-test was used. The chi-square or Fisher’s exact test was used to compare categorical variables on the basis of pre-specified criteria. The chi-square test was used to test 2 categorical variables when the expected frequencies were >5, and the Fisher’s exact test was used for expected frequencies of <5. Discrete variables are expressed as an absolute number (percentage). Missing data were not imputed and were assumed to be missing at random. A complete 1-year follow-up was available for all but 1 patient included in the analysis. Potential risk factors for 30-day mortality and recurrence were assessed using logistic regression. Cox regression was used to compare potential risk factors for survival up to 1 year with patients without event being censored at 1 year. After univariable analysis, all variables with a P-value of <0.1 were included in the multivariable model using forward selection (likelihood ratio, P_in = 0.05). Results are reported as odds ratios (OR) for 30-day mortality or hazard ratios (HR) for 1-year mortality with corresponding 95% confidence intervals (CIs) and P-values. To reduce the potential treatment selection bias in comparing the outcomes between groups, inverse probability weighting (IPW) was performed. Propensity scores were calculated through logistic regression analysis using the following variables: age, sex, diabetes, COPD, peripheral artery disease, chronic kidney disease, preoperative haemodialysis, septic embolism and S.aureus in blood culture. Inverse propensity scores were created and weighting for all patients was performed. For estimating causal effects, generalized estimation equation was performed. All P-values reported are two-sided. Statistical analyses were performed using SPSS Statistics version 28.

RESULTS

Characteristics of patients with infective endocarditis–spondylodiscitis coinfection

The IE study population included a total of 3991 patients. All patients had an indication for surgical treatment of the endocarditic valve. Concomitant SD occurred in 4.4% (n = 150) of IE patients. These patients with IE-SD coinfection were further analysed with regard to the sequence of therapy performed: in the ‘primary surgical treatment for IE’ group, cardiac surgery was performed 1st in 115 patients (76.6%). Subsequently, the spinal foci were treated either surgically (n = 31) or conservatively (n = 84). In the ‘primary surgical treatment for SD’ group 23.3% (35 patients), SD was 1st treated by spinal surgery and patients were subsequently operated on IE (Fig. 1).

Patients with IE-SD coinfection had a median age of 70.0 (64.0–75.6) years and were mostly male (79.5%). It is noticeable that patients with IE-SD coinfection had hypertension in 74.2%. In addition, diabetes mellitus was common in 33.1% and preoperative renal insufficiency in 47.7%.

Furthermore, there were no relevant differences in the manifestation of IE or SD: thus, the aortic valve was most frequently affected in both groups, followed by the mitral valve. Prosthetic endocarditis was present in 26.5%. Regarding the spine, SD occurred most frequently in the lumbosacral region in 58.3% followed by the thoracic region in 19.2%. Here, there were no relevant differences between the groups. Patients who underwent cardiac surgery 1st had received spinal infiltrations significantly more often beforehand (18.1% vs 3.2%, P = 0.020) (Table 1).

Looking at the bacterial spectrum of patients with IE-SD coinfection, blood cultures showed the most frequent detection of enterococci with 24.5%, followed by staphylococci with 21.2%, streptococci with 16.6% and coagulase-negative Staphylococci with 10.6% (Fig. 2). Patients who underwent surgical treatment of the SD 1st had a significantly higher incidence of S.aureus in blood cultures (40.6% vs 18.6%; P = 0.011) and in disc tissue (40.0% vs 15.8%; P = 0.017) (Table 1).

Clinical outcomes

In patients with IE-SD coinfection, 30-day mortality was 14.6% and 1-year mortality was 23.8%. If the spinal focus was operated on 1st, there was a significantly higher 30-day mortality compared to patients who underwent valve surgery 1st due to IE (25.7% vs 11.4%; P = 0.037) and a tendency for a higher 1-year mortality (34.3% vs 21.1%; P = 0.110). Patients initially treated surgically for IE had a higher recurrence rate of IE and/or SD during follow-up compared to patients initially treated surgically for SD (12.2% vs 0%; P = 0.023) (Table 2).

Independent predictors of mortality and recurrence

Our multivariable analysis revealed that primary surgical therapy for SD was an independent predictor of 30-day mortality (OR 3.816, 95% CI 1.849–7.876; P < 0.001). Furthermore, diabetes was an independent predictor of 30-day mortality (OR 3.927, 95% CI 1.697–9.085; P = 0.001). For 1-year mortality, COPD (HR 3.050; 95% CI 1.377–6.759; P = 0.006), preoperative haemodialysis (HR 3.493; 95% CI 1.591–7.669; P = 0.002) and a body mass index >25 (HR 2.864; 95% CI 1.068–7.683; P = 0.037) were shown to be independent predictors (Table 3).

Preoperative chronic kidney disease (OR 10.765, 95% CI 1.362–85.085; P = 0.024) as well as thoracic SD (OR 6.789, 95% CI 2.332–19.764; P < 0.001) were associated with recurrence of IE and/or SD (Table 4).

Even after a possible treatment selection bias was reduced when comparing the results between the groups using IPW, primary treatment of SD was found to be an independent risk factor for 30-day mortality [OR 3.513, CI (1.030–11.985); P = 0.045]. For 1-year mortality and recurrence rate of IE and/or SD, treatment sequence was not an independent risk factor after IPW.

DISCUSSION

We present an analysis of one of the largest patient cohorts with IE-SD coinfection to date, focusing on the influence of the sequence of surgical treatment (1st IE, then SD or vice versa). Our main findings were as follows: (i) patients who had surgery for SD 1st had higher 30-day mortality and a tendency for a higher 1-year mortality; (ii) recurrence rates were higher when IE was operated on 1st and (iii) primary surgical treatment of SD and diabetes were shown to be independent predictors of 30-day mortality, and COPD, preoperative haemodialysis and body mass index >25 were shown to be independent predictors of 1-year mortality. Recurrence was associated with preoperative chronic kidney disease and thoracic SD.

Prevalence and characteristics of infective endocarditis–spondylodiscitis coinfection

Carbone et al. report that in their IE collective, concomitant SD occurred in 8.5% [5]. This is consistent with Del Pace et al. who reported a prevalence of IE-SD coinfection in 8% of IE patients [4]. A higher prevalence of 19% was found in a smaller series of 58 IE patients by Tamura et al. [7]. Pigrau et al. showed that in an IE collective of 606 patients, concomitant SD was present in 4.6%. Conversely, a much higher coinfection rate was found in the SD collective, with concomitant IE detected in 30.8% of 91 SD patients [10]. In our multicentre IE collective with a total of 3991 patients, concomitant SD was detected in 4.4%. Compared to SD cohorts, the prevalence of IE-SD coinfection in IE cohorts is comparatively low. This may be due to the fact that IE patients complaining of back pain are often not systematically examined for SD, so that the prevalence of concomitant SD is often underestimated. In general, the detection of IE-SD coinfection is strongly influenced by the screening method used. In an SD collective, Behmanesh et al. showed that the rate of diagnosed IE increased 10-fold from 3% to 30% after the introduction of routine screening by transoesophageal echocardiography in patients with known SD [9].

Compared with other large IE collectives, patients with IE-SD coinfection are more likely to have secondary diseases and, at an average of 70 years, are slightly older than patients in other isolated IE collectives with a median age of 58–65 years [3, 17, 18]. In particular, we detected hypertension in three-quarters of patients with IE-SD coinfection, which is significantly more common than in patients with isolated IE [3]. Similar results were obtained by Carbone et al. who were able to show that patients with IE-SD coinfection were significantly older and, with 48% vs 34% (P < 0.001), had hypertension significantly more often than patients with isolated IE [5].

Looking at the bacterial spectrum, the detection of almost a quarter enterococci in our patients with IE-SD coinfection is striking compared with isolated IE collectives, in which enterococci make up a significantly lower proportion [3, 18]. In the CAMPAIGN IE cohort with 4300 IE patients, for example, 23% staphylococci, 15% streptococci and only 10% enterococci were detected [3]. This is in line with Carbone et al. who detected enterococci as the causative pathogen significantly more frequently in 24% of patients with IE-SD compared to 12% with isolated IE (P < 0.001) [5]. Del Pace et al. showed similar results: here, enterococci were detected in 35% of patients with IE-SD vs 16% of patients with isolated IE (P = 0.015) [4]. In addition, Del Pace et al. were able to demonstrate by means of multivariable analysis that the detection of enterococci was independently associated with the simultaneous occurrence of SD in IE patients [4].

Also, when looking at SD collectives, the most common pathogen detected in patients with isolated SD appears to be S.aureus [19, 20]. Yagdiran et al. showed that IE was significantly more frequent in the group of patients with enterococci and streptococci compared to S.aureus (26% vs 6%; P = 0.002) [19]. Similar results were obtained by Mullemann et al. whose retrospective study confirmed the high prevalence of IE-SD in streptococcal enterococcal infections [21]. Since enterococci occur to a considerable extent as causative agents of IE-SD coinfections, a clinical examination and, if necessary, a further diagnostic magnetic resonance imaging should be performed, especially in IE patients with evidence of enterococci, even if only mild symptoms such as back pain are present, in order to avoid overlooking or delaying the diagnosis of concomitant SD.

Mortality and recurrence in infective endocarditis–spondylodiscitis coinfection

Several studies have analysed whether IE-SD coinfection is associated with higher mortality than isolated SD or IE. Compared to patients with isolated SD, Viezens et al. [1] and Tamura et al. [7] found no difference in mortality in patients with IE-SD. On the other hand, in a large retrospective Japanese study, the presence of IE in patients with SD was associated with significantly higher in-hospital mortality (OR: 3.19, P < 0.001) [22]. In addition, Koslow et al. reported higher all-cause mortality in patients with IE-SD [23].

Looking at IE collectives, Carbone et al. found no difference in in-hospital and 1-year mortality in IE-SD patients compared to IE patients [5]. Del Pace et al. also showed that the presence of SD did not affect the short- or long-term prognosis of IE patients [4]. However, Del Pace et al. found significantly more recurrences in patients with concomitant SD, which they attributed to the high proportion of drug users in this subgroup [4].

Our data suggest that primary surgical therapy for SD is an independent predictor of 30-day mortality: patients who underwent spinal surgery 1st had a significantly higher 30-day mortality of 25.7% compared with patients who underwent cardiac surgery 1st (11.4%, P = 0.037). There were no differences in 1-year mortality depending on the treatment regime. Carbone et al. were also able to show that the presence of concomitant SD is not associated with a worse long-term prognosis if the diagnosis is made early and appropriate and prolonged antibiotic therapy is given [5].

Postoperative recurrence of IE or SD occurred more frequently in patients who underwent primary cardiac surgery compared with patients who initially underwent spinal surgery (12.2% vs 0%; P = 0.023). In our patients, this may be due to the fact that of the patients who underwent cardiac surgery 1st, only some subsequently underwent surgical therapy for SD and in many patients SD was only treated with antibiotics. Hence, it could be possible that the spinal focus was not treated sufficiently or that the antibiotic therapy was not carried out long enough.

Limitations

Our study has several limitations that should be considered when interpreting the results. First, our cohort might be influenced by referral bias because all the facilities were tertiary care centres. Therefore, the results of the present study may not be generalizable. There could be a possible bias due to centre-specific experience and routine use of imaging to evaluate an SD. Heterogeneity in defining variables, regional referral and practice patterns, and data collection methods could bias our results, as is the case in all multicentre studies. Due to the retrospective design and the multicentre approach, there is a lack of data in both groups. Since missing data occur comparatively frequently in both groups and in particular the outcome variables 30-day mortality and 1-year mortality are present in all but 1 patient, we assume that the data are missing by chance and have no relevant influence on the analysis. As this is a non-randomized study, selection bias may be present, potentially influencing the prognosis. Therefore, no causality can be drawn between the sequence of therapy and the result. Furthermore, the patients were often treated in peripheral hospitals before they were transferred to our maximum care centre and after they were operated on, the exact start and duration of anti-infective therapy could not always be determined. Although this is a large population for this specific subset of IE patients, it remains underpowered to detect significant differences between management strategies and groups and the number of cases may be too small to reliably assess the prevalence of recurrence.

Nonetheless, we report one of the largest cohorts of patients with IE-SD coinfection, and we hope that our data can contribute to a better knowledge of patient characteristics and postoperative outcomes in these complex patients. However, prospective, randomized data would be desirable and interesting in order to be able to make a meaningful recommendation on the treatment sequence without the limiting selection bias that occurs in retrospective studies such as this one.

CONCLUSION

Our data show that primary surgical treatment of SD in patients with IE-SD coinfection is associated with increased 30-day mortality and is an independent risk factor for 30-day mortality. Primary surgical treatment of IE is associated with more recurrences of IE and/or SD. Overall, if clinically suspected or if certain risk factors are present, such as a pathogen detection with enterococci, further imaging should be performed early to rule out concomitant SD in IE patients.

FUNDING

No funding was received.

Conflict of interest: none declared.

DATA AVAILABILITY

The data underlying this article will be shared on reasonable request to the corresponding author.

Author contributions

Carolyn Weber: Conceptualization; Data curation; Formal analysis; Methodology; Project administration; Validation; Writing—original draft. Martin Misfeld: Conceptualization; Data curation; Investigation; Supervision; Validation; Writing—review & editing. Mahmoud Diab: Conceptualization; Data curation; Investigation; Validation; Writing—review & editing. Shekhar Saha: Conceptualization; Data curation; Investigation; Validation; Writing—review & editing. Ahmed Elderia: Conceptualization; Data curation; Investigation; Validation; Writing—review & editing. Mateo Marin-Cuartas: Data curation; Project administration; Validation; Writing—review & editing. Maximilian Luehr: Conceptualization; Data curation; Investigation; Validation; Writing—review & editing. Ayla Yagdiran: Conceptualization; Data curation; Investigation; Validation; Writing—review & editing. Peer Eysel: Conceptualization; Investigation; Supervision; Validation; Writing—review & editing. Norma Jung: Conceptualization; Investigation; Supervision; Validation; Writing—review & editing. Christian Hagl: Conceptualization; Investigation; Supervision; Validation; Writing—review & editing. Torsten Doenst: Conceptualization; Investigation; Supervision; Validation; Writing—review & editing. Michael A. Borger: Conceptualization; Investigation; Supervision; Validation; Writing—review & editing. Nikolaus Kernich: Conceptualization; Data curation; Investigation; Validation; Writing—original draft. Thorsten Wahlers: Conceptualization; Investigation; Supervision; Validation; Writing—original draft.

Reviewer information

European Journal of Cardio-Thoracic Surgery thanks Ari Mennander, Jose Lopez-Menendez and Patrick Myers for their contribution to the peer review process of this article.

Presented at the European Association of Cardio-Thoracic Surgery’s 37th Annual Meeting, Vienna, Austria, 5th October 2023.

REFERENCES

ABBREVIATIONS

ABBREVIATIONS
 
• CI

  Confidence interval

 
• COPD

  Chronic obstructive pulmonary disease

 
• ESC

  European Society of Cardiology

 
• HR

  Hazard ratio

 
• IE

  Infective endocarditis

 
• IE-SD

  Infective endocarditis–spondylodiscitis

 
• MRI

  Magnetic resonance imaging

 
• OR

  Odds ratio

 
• SD

  Spondylodiscitis

Author notes