Prognostic value of pro-adrenomedullin and copeptin in acute infective endocarditis

This was a retrospective study conducted at the Unit of Infectious & Transplant Medicine, Monaldi Hospital, University of Campania “Luigi Vanvitelli”, involving patients with a diagnosis of IE admitted between 2007 and 2019 and for whom a plasma sample obtained in the acute phase of the disease was available. IE diagnosis was done according to existing criteria over time (modified Duke criteria until 2014 and ESC criteria from 2015 on) [25, 26]. All patients hospitalized at our Unit with diagnosis of IE undergo blood sampling. In the period 2007–2019 we observed in our Hospital, a regional referral center for IE, 525 IE cases. We included in this study all adult patients (n = 196) who were admitted to our Unit with a recent diagnosis of IE and without need of emergent surgery. They were referred from other Departments of our hospital (4 Cardiology units), other hospitals and after observation as outpatients. Most referred patients (n = 327) were excluded, although we had a sample available, because this sample was a later sample during the course of the disease. No patient with shock or septic shock was included, as these are routinely admitted to the Intensive Care Unit (ICU), not our Unit.

The study was approved by the Ethics Committee of the University of Campania “Luigi Vanvitelli” and AORN Ospedali dei Colli. All patients gave their written informed consent to blood sampling and the anonymous use of their clinical data.

Detailed data of patients were available as part of a standardized protocol of IE evaluation in use at our Unit, which includes a baseline clinical evaluation, particularly clinical history, physical examination, body mass index (BMI), chest X-ray, abdominal ultrasound scan and laboratory analyses (including CRP, creatinine, urea, glycemia, blood count, INR). According to the protocol, a trans-thoracic echocardiogram (TTE) was performed in all patients within 72 h of admission, followed by a trans-esophageal echocardiogram (TEE) where needed. Detailed information about IE characteristics (on native, prosthetic or cardiac implantable electronic device [CIED]), endocardial vegetations (number, size and position) and isolated causative pathogens were also collected. Embolic events, defined as acute complications causing overt clinical manifestations [27], and their characteristics (location, extension, complications) were also recorded.

Chronic heart failure (CHF), chronic kidney disease (CKD), liver disease and diabetes mellitus were considered as the principal co-morbidities. The Charlson comorbidity index (CCI) was also calculated for each patient.

PCT, pro-ADM and copeptin were measured in included patients on plasma samples, as detailed below. The obtained values of these biomarkers were analysed in relation with clinical (age, sex, embolism, involved valves, co-morbidities) and laboratory (CRP, white blood cells, creatinine, glycemia, alanine transferase (ALT), isolated microorganism) parameters. The estimated glomerular filtration rate (eGFR) was calculated by Modification of Diet in Renal Disease (MDRD) formula. In addition, PCT, pro-ADM and copeptin plasma levels were studied in relation to short term (at hospital discharge) and long term (at 1 year from the IE diagnosis) mortality.

Blood was collected in EDTA tubes (Greiner Bio-One, Kremsmünster, Austria), cooled at 4 °C and centrifuged; plasma samples were stored at − 80 °C at the time of IE diagnosis, for subsequent use.

Plasma concentration of pro-ADM (cut-off for positivity 0.38 nmol/L), copeptin (cut-off for positivity 3.9 pmol/L) and PCT (cut-off for positivity 0.064 μg/L) were measured using an automated immunofluorescent assay on a Kryptor system (B.R.A.H.M.S. AG, Henningsdorf, Germany).

Other laboratory parameters were obtained by routine methods used in our Hospital central laboratory, including C-reactive protein (cut-off for positivity 0.3 mg/dL).

Numerical data are presented as median with range, whilst categorical/nominal data as number and percentage. The Mann-Whitney U test was used to assess statistical significance of the differences between numerical groups of variables and the Fisher’s exact test for nominal variables. Logistic regression analysis of independent predictors of hospital mortality and 1-year mortality was performed by block entering in the model all variables significantly associated with each of these outcomes on the univariate analysis. Correlation between numerical variables was assessed by Spearman’s coefficient. The biomarkers pro-ADM, copeptin, PCT and CRP have been studied altogether for the first time in IE and no specific cut-off correlated to short and long term outcome of the disease was available. Using as positivity cut-offs the biomarker values indicated on the manufacturer instruction manual as reported in the methods section, most patients had positive values (pro ADM 185 vs 11, PCT 169 vs 27, copeptin 177 vs 19, CRP 187 vs 9). Accordingly, we evaluated each biomarker using as cut-off its median value observed in the overall study group.

To assess the predictive performance of biomarker levels on IE outcome, we calculated the area under the Receiver Operating Characteristic (ROC) curve, entering in-hospital mortality as the state variable. The significance level was set at 5% and all tests were 2-tailed. All analyses were performed using the statistical software for Windows SPSS 20 (SPSS, Inc., Chicago, Illinois, USA).

Biomarker levels were evaluated in relation to major clinical features and hospital mortality of IE patients (Table 2). At univariate analysis, pro-ADM and copeptin were significantly higher in older patients. In-hospital mortality was significantly associated with higher levels of pro-ADM, PCT and copeptin, while 1 year mortality was significantly associated only with higher levels of pro-ADM and copeptin (Table 2).

We subsequently performed a multivariate logistic regression analysis to identify variables independently associated to IE outcomes. As shown in Table 3, panel A, only pro-ADM was an independent predictor of in hospital mortality (OR 3.29 [95%C.I. 1.04–11.5]; p = 0.042) (Table 3). In contrast, higher copeptin levels were independently associated to 1 year mortality (OR 2.55 [95%C.I. 1.18–5.54]; p = 0.017), while pro-ADM showed a trend for an association with this outcome (OR 1.93 [95%C.I. 0.9–4.12]; p = 0.088) (Table 3, panel B).

At ROC curve analysis, pro-ADM and copeptin appeared to best predict in-hospital mortality (areas under the ROC curves: 0.784 and 0.731, respectively) compared with PCT and CRP (Fig. 1).

Embolic events occurred in 77 patients (39.3%). CRP was the only biomarker associated with embolic events (p = 0.003) (Table 2). Most patients (151, 77%) had an indication for cardiac surgery at completion of diagnostic assessment, but this outcome was not related to any of the studied biomarkers (Table 2). Of these 151 patients, 128 (84.7%) finally underwent surgery, but actual surgical treatment was not associated with any biomarker level (Table 2).

Pro-ADM levels were higher in patients with left-sided IE compared to those with right-sided IE (p = 0.001; see Additional Table 2a), while no differences were observed for any biomarker in relation to the actual subtype of infection (native valve vs prosthetic valve vs CIED, see Additional Table 2b).

To evaluate association with IE etiology, we analyzed biomarker levels according to IE causative pathogen (Fig. 2). Interestingly, higher levels of pro-ADM were observed in patients with infection due to Enterococci and Staphylococci (Fig. 2a), whereas Staphylococcus aureus cases had the highest PCT levels (Fig. 2b). S. aureus IE cases also had higher CRP levels (Fig. 2d). The enterococcal etiology was associated to a significantly higher copeptin level (Fig. 2c). Streptococcal IE was characterized by the lowest biomarker levels (Figs. 2a-d2).

Table 4 summarizes the relationship between biomarkers and co-morbidities. Liver disease and diabetes mellitus did not influence biomarker plasma concentration (Table 4). High levels of pro-ADM were significantly influenced by prior CHF and CKD while high levels of copeptin were associated with CKD only. CRP was also higher in the presence of CHF (p = 0.05). As shown in Additional Fig. 1, the CCI showed a direct correlation with pro-ADM and copeptin, but not CRP or PCT.

We also assessed whether an IE-related kidney dysfunction influenced biomarker levels. As shown in Table 4, all biomarkers were significantly higher in patients with a baseline eGFR < 60 ml/min, except CRP.

Comorbidities were also associated with clinical outcomes (Additional Table 3). In particular, kidney disease and diabetes were associated with hospital mortality, whereas heart failure and kidney disease were related with 1 year mortality. Also, the CCI was significantly associated with hospital mortality (median 4 [0–13] in deceased vs 3 [0–13] in survivors; p = 0.023) as well as 1 year mortality (median 5 [0–13] in deceased vs 3 [0–13] in survivors; p < 0.001).

To the best of our knowledge, this is the first study evaluating pro-ADM and copeptin in IE. As previously observed in other infectious syndromes [10, 13, 18, 19], pro-ADM and copeptin were strong and independent predictors of hospital and 1 year mortality, respectively, in IE. Pro-ADM levels showed a nearly 80% sensitivity for hospital death. In contrast, PCT did not predict IE mortality.

PCT and CRP were tied to other disease features, including a staphylococcal etiology for PCT and the occurrence of embolic events for CRP.

Our data fill the knowledge gap on the role of PCT in IE, where controversial findings emerged in previous studies. PCT previously appeared a useful biomarker for the diagnosis of IE, and also showed a correlation with specific bacterial etiology and disease prognosis [22‐24]. However, other studies failed to demonstrate superiority of PCT over CRP in predicting IE [23]. In our study, PCT showed a pattern similar, but not overlapping with CRP, but none had an independent predictive role for IE mortality. Only higher CRP concentrations confirmed to be associated with embolic events [27], a serious complication in IE.

Our results on the relation between biomarkers and IE etiology deserve comment. Different IE etiology may be associated with diverse risk factors, disease course, treatment response and time of referral. It was therefore interesting to observe how copeptin and pro-ADM, strongly increased in older patients, were associated with enterococcal IE, that typically occurs with increasing incidence in the multi-morbid elderly [28]. In agreement with other studies [22‐24], we found that PCT levels were highest in infections due to S. aureus or gram negative bacteria.

As many patients with IE have co-morbidities, it was interesting to evaluate their effect on/relation with biomarker levels. Pro-ADM and copeptin were higher in IE patients with a history of CKD and CHF, confirming the association of these biomarkers with organ failure [10, 13, 18, 19]. Moreover, pro-ADM and copeptin correlated with both NT-pro-BNP and creatinine, suggesting a relation with acute, IE-associated cardiac dysfunction, and also a strong influence of renal dysfunction on biomarker levels.

As mentioned, older age translated into higher pro-ADM and copeptin, and this could influence their effect on mortality. However, these biomarkers predicted mortality (either hospital or 1 year) independent of age.

This was a single center, retrospective study and determination of the biomarkers pro-ADM, PCT and copeptin was performed on stored samples at a single time-point. The study sample was relatively small, although homogeneously managed according to existing guidelines.

Previous conflicting data suggested a key role for PCT in the assessment of antimicrobial therapy effectiveness in infectious syndromes other than IE [7‐9]. Unfortunately, we could not analyze the on treatment dynamics of PCT to predict therapeutic response in this study.

Finally, we were unable to confirm our findings in a prospective validation cohort.