In this study, we reviewed the clinical, laboratory, and echocardiographic data of patients with infective endocarditis associated with ARA. We reported several clinical (e.g., mechanical prosthesis), laboratory (e.g., higher CRP), and echocardiographic (e.g., aortic vegetations) variables that were significantly higher in ARA patients. The odds of ARA were higher in patients with a mechanical prosthesis, aortic vegetations, aortic paravalvular leakage, and undetected organisms and lower in those with mitral valve vegetations.
The prevalence of ARA was variable in past studies. An earlier report by John et al. detected ARA in 32 out of 50 patients with aortic IE (46%) from 1982 to 1988 [6]. In Leung et al., ARA was detected in 32% of patients (11 out of 34 patients with aortic IE) in the period from 1989 to 1993 [17]. Anguera et al. identified ARA in 201 patients out of 2055 native aortic valve IE (9.8%) in a retrospective analysis from 16 referral hospitals from 1992 to 2003 [18]. In the same issue, Anguera et al. reported ARA in 150 patients out of more than 872 prosthetic aortic valve IE (17%) from the same registry [19]. Graupner et al. detected a higher prevalence of ARA in 37% of aortic IE patients (78 out of 211 patients from 1996 to 2000). In the TAVI era, prosthetic valve endocarditis was detected in 103 out of 4336 patients. Among these 103 patients, ARA was detected in 12 patients (11.9%) [20]. Our study reported ARA in 21.4% of patients with left-sided IE and 40.1% of aortic valve IE. Late presentation and use of inappropriate antibiotics could explain the high percentage in this report, which reflects the current situation in Egypt.
The median age of the study population was 31.5 years (IQR; 24, 42.2), which is much less than the worldwide reported age of ARA patients. In a recent meta-analysis, the age of ARA patients ranged from 37 to 62 years [21]. The young age of our patients reflects the type of the underlying heart disease in our study, being rheumatic or congenital heart disease in the majority of cases and degenerative heart disease in only 9.5% of cases. Age was not statistically different between both groups (ARA and NO-ARA groups), but the male gender was more common in the ARA group. Other reports for aortic root abscess also showed a trend of more incidence of ARA in male patients [9, 21].
Comorbid conditions such as diabetes mellitus and renal and hepatic diseases were similar in both groups. There was a trend of higher chronic steroid use in the NO-ARA group. Most of the patients on steroid therapy (11 patients) had mitral valve endocarditis, and only one patient had aortic and mitral valve endocarditis without abscess development. However, it is improper to assume any protective role of steroids from aortic root abscess.
We found that prosthetic mechanical valve and congenital heart disease were significantly higher in the ARA group. The prosthesis was found to be an independent predictor for ARA in previous studies [22]. The bicuspid aortic valve was the most common congenital heart disease in our study population and was significantly higher in the ARA group. Kiyota et al. [23] has shown that the bicuspid aortic valve is associated with an increased incidence of IE and ARA as compared with the trileaflet valve.
Clubbing was seen more in the NO-ARA group. This could be explained by the fact that clubbing needs some time to develop, which is not the case for IE with ARA, which is usually an aggressive, rapidly developing infection. CRP levels were higher in the ARA group. This finding could be explained by the more extensive damage and the more aggressive nature of ARA. However, we did not find any previous reports describing the above two findings.
There was an increased likelihood of microorganisms’ un-detection in our study population (i.e., negative blood cultures and serology) that may be attributed to increased use of antibiotics before referral. This microorganism un-detection was numerically higher in the ARA group compared with the No-ARA group (54.1% vs. 42%, p = 0.09). On the other hand, the detected microorganisms were not different between both groups. Staphylococcus aureus and fungal infections represent the most commonly detected organisms in the ARA group (each occurred in 11.5% of ARA patients). Staphylococcus aureus was commonly found in ARA patients in previous studies [24, 25].
Regarding the complications, there was no statistically significant difference between both groups, including all-cause in-hospital mortality. These results contrast previous data that showed a worse outcome of IE complicated with aortic root abscess [26, 27]. The relatively late presentation and delayed surgical intervention in both groups could be the cause of the comparable outcome.
In our study, the ARA group showed more aortic valve vegetations and less mitral valve vegetations. This finding seems reasonable in a study of aortic root abscess and not mitral ring abscess. However, this may raise a question regarding the well-known site for aortic root abscess, which is the aortomitral continuity. Forteza et al. operated upon 26 patients with aortic valve IE and intervalvular fibrous body abscess representing 10.6% of the patients with aortic valve IE [28]. Our study suggests that the spread from aortic valve endocarditis is the original site of the ARA that can extend into aortomitral continuity, rather than perivalvular infection in mitral valve endocarditis.
The paravalvular leak was significantly higher in the ARA group. The paravalvular leak is a common finding in prosthetic valve endocarditis. Anguera et al. detected moderate to severe aortic paravalvular leak in 45% of patients with prosthetic valve IE. Prosthetic valve IE usually begins as periannulitis and then spreads to adjacent tissues causing an abscess and can lead to a paravalvular leak [19]. The presence of moderate to severe regurgitant valve lesion was found to be lower in the ARA as compared to the NO-ARA group. This result can be explained in two aspects. First, all implanted prostheses in our study population were mechanical, where paravalvular leakage is much more common than transvalvular regurgitation. Second, perhaps the original site of the infection in ARA is more eccentric at the annulus more than the valve itself, favoring the spread of infection to adjacent tissues causing periannular complications rather than destroying the valve itself.
Multivariate analysis of the above variables showed that the presence of mechanical prosthesis and paravalvular leakage were the most independent predictors of ARA (OR 3.7 and 3.9, respectively). The presence of mechanical prosthesis also seems to be associated with an increased risk of development of ARA. This result highlights the importance of appropriate perioperative sterilization and disinfection, as well as the necessity of proper hygiene in patients with prosthetic valves. The presence of paravalvular leakage of any degree in the presence of clinical suspicion should be alarming to the possibility of ARA. All required investigations such as transesophageal echocardiography and CT aortography should be done to exclude ARA. Other independent predictors were aortic valve vegetations and undetected organisms. Both variables suggest the presence of the aggressive nature of infection that leads to this severe annular complication. ARA was less seen in the presence of mitral valve vegetations.
Studies reporting the predictors of ARA are scarce and relatively old. Omari et al. reported aortic valve infection and intravenous drug abuse as the most independent predictors of ARA in patients with native aortic valve IE [24]. Later, Blumberg et al. identified a new atrioventricular or bundle branch block as the only significant correlation [29]. In another study, the most common risk factors for paravalvular infection were prosthetic valve, aortic valve infection, and coagulase-negative staphylococci [30]. We could not find any recent studies highlighting the ARA predictors.
The latest trials of ARA focused on the outcome of surgical procedures. A meta-analysis by Chen et al. [21] reviewed seven surgical trials of ARA, comparing the results of aortic root replacement vs. aortic valve replacement. There was no difference between both procedures on 30-day follow-up; however, aortic root replacement was associated with a 50% reduction of the rate of reoperation on 1-year follow-up. Kirali et al. [31] showed surgical outcomes in 27 patients with ARA. The mean duration of follow-up was 6.8 ± 3.7 years. In-hospital mortality was 22.2%, which was lower than in-hospital mortality reported in our study (36.1%). Mean 1-, 5-, and 10-year survival were 70.2%, 62.2%, and 59.2%, respectively. Sultan et al. [32] studied the use of aortic homograft in 138 ARA patients with relatively not high surgical mortality (12.3%). However, 5-year mortality was again high (43%). Yang et al. [33] showed similar results in 179 patients with an operative mortality of 8.4% and 10-year mortality of 59%. So, despite the good immediate surgical outcome in ARA patients, this complication carries an increased risk of mortality on long-term follow-up.
There are some limitations to our study. First, it was an observational study, being limited by the lack of local resources and expertise. Second, underlying cardiovascular conditions, time to presentation, the causative microorganisms, and antibiotic regimen protocols may differ between different countries. Third, the relatively small number of ARA patients is considered a limitation. However, the majority of previous studies reported their results based on a similar number of patients due to the slow recruitment of ARA in clinical studies. Finally, the significant variables by multivariate analysis in our study showed a wide confidence interval. The leading cause of this wide CI was the limited sample size. The confidence intervals of odds ratios were also wide in the previous ARA studies. In Leontyev et al., one of the largest studies of surgical treatment of ARA with 172 patients, all independent predictors of mortality showed wide CI (e.g., sepsis had OR 3.6 with 95% CI 1.2–10.7) [34]. Increasing sample size could lead to a narrower confidence interval; however, the recruitment of a large number of ARA patients into a clinical study still represents a challenge.