Circulating adrenomedullin estimates survival and reversibility of organ failure in sepsis: the prospective observational multinational Adrenomedullin and Outcome in Sepsis and Septic Shock-1 (AdrenOSS-1) study

Background Adrenomedullin (ADM) regulates vascular tone and endothelial permeability during sepsis. Levels of circulating biologically active ADM (bio-ADM) show an inverse relationship with blood pressure and a direct relationship with vasopressor requirement. In the present prospective observational multinational Adrenomedullin and Outcome in Sepsis and Septic Shock 1 (, AdrenOSS-1) study, we assessed relationships between circulating bio-ADM during the initial intensive care unit (ICU) stay and short-term outcome in order to eventually design a biomarker-guided randomized controlled trial. Methods AdrenOSS-1 was a prospective observational multinational study. The primary outcome was 28-day mortality. Secondary outcomes included organ failure as defined by Sequential Organ Failure Assessment (SOFA) score, organ support with focus on vasopressor/inotropic use, and need for renal replacement therapy. AdrenOSS-1 included 583 patients admitted to the ICU with sepsis or septic shock. Results Circulating bio-ADM levels were measured upon admission and at day 2. Median bio-ADM concentration upon admission was 80.5 pg/ml [IQR 41.5–148.1 pg/ml]. Initial SOFA score was 7 [IQR 5–10], and 28-day mortality was 22%. We found marked associations between bio-ADM upon admission and 28-day mortality (unadjusted standardized HR 2.3 [CI 1.9–2.9]; adjusted HR 1.6 [CI 1.1–2.5]) and between bio-ADM levels and SOFA score (p < 0.0001). Need of vasopressor/inotrope, renal replacement therapy, and positive fluid balance were more prevalent in patients with a bio-ADM > 70 pg/ml upon admission than in those with bio-ADM ≤ 70 pg/ml. In patients with bio-ADM > 70 pg/ml upon admission, decrease in bio-ADM below 70 pg/ml at day 2 was associated with recovery of organ function at day 7 and better 28-day outcome (9.5% mortality). By contrast, persistently elevated bio-ADM at day 2 was associated with prolonged organ dysfunction and high 28-day mortality (38.1% mortality, HR 4.9, 95% CI 2.5–9.8). Conclusions AdrenOSS-1 shows that early levels and rapid changes in bio-ADM estimate short-term outcome in sepsis and septic shock. These data are the backbone of the design of the biomarker-guided AdrenOSS-2 trial. Trial registration ClinicalTrials.gov, NCT02393781. Registered on March 19, 2015. Electronic supplementary material The online version of this article (10.1186/s13054-018-2243-2) contains supplementary material, which is available to authorized users.


Introduction
Adrenomedullin (ADM) is a free circulating peptide with potent vascular properties, including benefits for endothelial barriers at physiological levels. ADM has previously been described as a "double-edged sword" in sepsis [1] because high levels of ADM induce vasodilation and hypotension [2][3][4] on one hand while reinforcing the endothelial barrier and improving outcome on the other [5][6][7][8][9][10]. The potential of ADM as a prognostic biomarker has previously been studied in critically ill patients, often by measuring the inactive midregional pro-ADM [11,12], or recently by direct measurement of the bioactive form of ADM (bio-ADM) [13,14]. It has been shown repeatedly that bio-ADM greater than 70 pg/ml is associated with worse outcome [13,14].
On the basis of previous results, we tested the hypothesis that modulating the ADM pathway in patients with high levels of circulating bio-ADM may improve shortterm outcome in sepsis. Adrecizumab, a monoclonal anti-ADM antibody, has been shown to improve organ function in preclinical settings [15]. In order to design a human trial in which we would administer adrecizumab based on levels of bio-ADM, we needed to assess the relationship between initial levels of bio-ADM and short-term outcome in sepsis and in septic shock patients.
In the Adrenomedullin and Outcome in Sepsis and Septic Shock 1 (AdrenOSS-1) study, we investigated whether the initial plasma concentration of bio-ADM (on intensive care unit [ICU] admission and after 48 h) may provide insight into 28-day survival and the recovery of organ function.

Study design
AdrenOSS-1 was a European prospective observational study. Twenty-four centers in five countries (France, Belgium, The Netherlands, Italy, and Germany) contributed to the trial achievement of 583 enrolled patients. Patients were recruited from June 2015 to May 2016. The study protocol was approved by the local ethics committees and was conducted in accordance with Directive 2001/20/EC, as well as good clinical practice (International Conference on Harmonization Harmonized Tripartite Guideline version 4 of May 1, 1996, and decision of November 24, 2006) and the Declaration of Helsinki.
The study enrolled patients aged 18 years and older who were (1) admitted to the ICU for sepsis or septic shock or (2) transferred from another ICU in the state of sepsis and septic shock within less than 24 h after admission. Included patients were stratified by severe sepsis and septic shock based on definitions for sepsis and organ failure from 2001 [16]. In the present article, the term "sepsis" refers to the updated definition of Sepsis-3 [17]. Concerning septic shock, most data presented in this article are based on the former definition [16], except for the confirmatory analyses presented in the last paragraph of the "Results" section, for which the new Sepsis-3 definition of septic shock was used [17].
Exclusion criteria were pregnancy, vegetative coma, and participation in an interventional trial in the preceding month. Informed consent was obtained from all patients or their lawful representatives prior to enrollment in the study. Patients were treated according to local practice, and treatments as well as procedures were registered.
The primary endpoint was 28-day mortality. Secondary endpoints concerned organ failure (as defined by the Sequential Organ Failure Assessment [SOFA] score) and organ support, vasopressor/inotrope use, fluid balance, and use of renal replacement therapy (RRT), as well as validation of the previously identified cutoff value of 70 pg/ ml [14]. The latter was identified as the optimal screening cutoff for AdrenOSS-2, an ongoing proof-of-concept and dose-finding phase II trial assessing adrecizumab (an antibody modulating circulating bio-ADM) in patients with early septic shock (NCT03085758). The relationship between cardiovascular SOFA subscore and bio-ADM, being a biomarker of vascular dysfunction, was evaluated.

Collection of patient data
Upon admission, demographics (age, sex), body mass index, presence of septic shock, type of ICU admission, organ dysfunction scores (SOFA, Acute Physiologic Assessment and Chronic Health Evaluation II [APACHE II]), origin of sepsis, preexisting comorbidities (i.e., treated within the last year), past medical history, laboratory values, and organ support were recorded, and blood was drawn for measurement of bio-ADM and other markers.
After patient enrollment, the following data were collected daily during the first week: SOFA score, antimicrobial therapies, fluid balance, ventilation status, Glasgow Coma Scale score, central venous pressure, need for RRT, invasive procedures for sepsis control, and vasopressor/inotrope treatment. Moreover, discharge status and mortality were recorded on day 28 after ICU admission.

Sample collection
Blood for the central laboratory was sampled within 24 h after ICU admission and on day 2 (mean 47 h, SD 9 h) after the first sample. Samples were subsequently processed and stored at − 80°C before transfer to the central laboratory for blinded bio-ADM analysis organized by the study sponsor (sphingotec GmbH, Hennigsdorf, Germany). Routine analyses (e.g., partial pressure of arterial oxygen, lactate) were performed by the local laboratories.

Bio-ADM measurement
Bio-ADM was measured using a recently developed immunoassay provided by sphingotec GmbH. For details and design principles on the assay, see publications by Marino et al. [14] and Weber et al. [18]. The analytical assay sensitivity was 2 pg/ml.

Statistical analyses
Results are presented as number and percentage, mean and SD, or median and IQR, depending on their distribution. Group comparisons for continuous variables were performed using the Kruskal-Wallis test, and appropriate post hoc tests were applied if necessary. Categorical data were compared using the chi-square test with simulated p values using 2000 replicates. Biomarker data were log-transformed if necessary. Cox proportional hazards regression was used to analyze the effect of risk factors on survival in uni-and multivariable analyses. The assumptions of proportional hazards were tested for all variables. For continuous variables, HRs were standardized to describe the HR for a biomarker change of one IQR. CIs (95% CI) for risk factors and significance levels for chi-square (Wald) test are given. The predictive value of each model was assessed by the model likelihood ratio chi-square statistic. The concordance index (C index) is given as an effect measure. It is equivalent to the concept of AUC adopted for binary outcome. For multivariable models, a bootstrap-corrected version of the C index is given. To test for added predictive value, we used the likelihood ratio chi-square test for nested models to assess whether bio-ADM adds predictive value to a clinical model or a risk score. Survival curves plotted by the Kaplan-Meier method using quartiles or predefined cut points (70 pg/ml) of bio-ADM were used for illustrative purposes. ROC curve analysis was applied for 28-day mortality to determine the optimal Youden cutoff in this cohort.
A two-sided p value of 0.05 was considered statistically significant. All analyses were performed using R version 2.5.1 (http://www.r-project.org, library Design, Hmisc, ROCR) and IBM SPSS Statistics version 22.0 software (IBM, Armonk, NY, USA).

Results
A total of 583 patients were included in the AdrenOSS-1 study. Patient characteristics, organ dysfunction scores, physiological and laboratory values, organ support upon admission, and outcome parameters are presented in Table 1. The median bio-ADM level at admission was 80.5 pg/ml [IQR 41.6-148.1] in our studied patients; 55.9% had bio-ADM level greater than 70 pg/ml at admission, and 44.1% had a bio-ADM less than 70 pg/ml. Of note, patients with septic shock had a significantly higher bio-ADM concentration at admission than patients with sepsis (114.4 [62.6-214.5] versus 57.5 pg/ml [31.2-101.5], p < 0.0001).

Bio-ADM levels and mortality
Over the 28-day follow-up period, 127 patients (22%) died: 33 with sepsis and 94 with septic shock.
In a Cox proportional hazards model adjusted for age, gender, comorbidities (cardiac and noncardiac), lactate, and diagnosis (sepsis, septic shock), bio-ADM concentration at admission was independently associated with 28-day mortality in the studied population (added chi-square 12.2, p = 0.0005; adjusted standardized HR 1.6 [95% CI 1.1-2.5], p = 0.0004) ( Table 2). Noticeably, the C index for prediction of 28-day mortality for bio-ADM at admission was 0.688 (95% CI 0.642-0.733, chi-square 54.8, p < 0.0001) in the univariate Cox regression. C indexes for lactate, SOFA, and APACHE II were 0.720 (95% CI 0.672-0.768), 0.728 (95% CI 0.680-0.777), and 0.701 (95% CI 0.657-0.746), respectively (all p < With the predefined cutoff value of 70 pg/ml, Kaplan-Meier analysis confirmed predictive value of bio-ADM for 28-day mortality in all studied patients (Additional file 1: Figure S1) and in subgroups of sepsis and septic shock ( Fig. 1a and b). Patient characteristics for high and low bio-ADM levels are illustrated in Table 1, and characteristics for survivors versus nonsurvivors are provided in Additional file 2: Table S1. The optimal Youden cutoff in all patients was 101.9 pg/ml (sensitivity 67.7%, specificity 67.3%). In septic shock, the optimal Youden cutoff was 99.1 pg/ml (sensitivity 71.3%, specificity 52.3%), and in severe sepsis it was 101.9 pg/ml (sensitivity 57.6%, specificity 78.6%). This compares with a sensitivity of 77.2% and specificity of 48.9% in all patients for the predefined bio-ADM cutoff of 70 pg/ml.
We additionally assessed outcome in relation to bio-ADM changes in the initial 48 h in time-dependent  Figure S2) and was independent of time-dependent lactate or SOFA score evaluation (Table 2). Patients were divided into four groups based on baseline and day 2 bio-ADM concentrations and under implementation of the cutoff value of 70 pg/ml: remaining low (low-low, LL), high-tolow (HL), low-to-high (LH), and remaining high (high-high, HH). Patient characteristics of these subgroups are presented in Additional file 4: Table S2.
In patients admitted with high bio-ADM upon admission, those who decreased bio-ADM towards normal values within the first 48 h (HL group) had a similar 28-day mortality to the LL group (HL 9.5%, LL 10.5%) and a more favorable outcome than patients whose bio-ADM remained high (HH group) or became high (LH group) (28-day mortality of 38.1% and 38.2%) (Additional file 4: Table S2).

Bio-ADM levels and organ dysfunction
Bio-ADM levels upon admission correlated with the initial SOFA score in all studied patients (n = 509, r = 0.49, p < 0.0001) (Additional file 5: Figure S3). SOFA score was higher in patients in septic shock than in those in sepsis, and for each group in patients with high initial bio-ADM (Additional file 6: Figure S4). Figure 3a indicates that the initial level of circulating bio-ADM relates to the need for and duration of organ support in survivors (p < 0.0001).
Concerning circulating bio-ADM levels and cardiovascular function, we found an almost linear relationship of bio-ADM and both cardiovascular SOFA subscore (p <   Table S3). Our analysis further revealed that patients with high bio-ADM at admission needed more vasopressors/inotropes over the following 7 days even if they did not have those treatments at admission (Fig. 3c). Regarding other organ support, patients who needed volume resuscitation of more than 5 L over the first 2 days (Fig. 3d) (p < 0.0001) or RRT (Additional file 9: Figure S6) or had long ICU stay (Additional file 10: Figure S7) had much higher circulating bio-ADM levels upon ICU admission than those patients who did not.
In agreement with the fact that serial measurements of bio-ADM indicated survival benefit in patients who dropped bio-ADM levels at day 2, we could demonstrate that drop of bio-ADM over the first 2 days also preceded the decrease of total SOFA score (p value for differences between HH vs. HL: p < 0.0001 for all days) (Fig. 4).

Discussion
The AdrenOSS-1 study was a prospective multinational observational cohort study assessing the relationship between rapid changes in circulating bio-ADM levels in the first 2 days and clinical outcome in ICU patients with sepsis and septic shock. We confirmed elevated levels of bio-ADM in septic patients and the striking relationship between circulating bio-ADM at ICU admission, organ dysfunction, and death. We also demonstrated that early recovery of circulating bio-ADM levels towards normal values (i.e., < 70 pg/ml) was associated with normalization of vascular function and better 28-day survival.
Our study found moderately elevated circulating levels of bio-ADM at admission in sepsis and strongly elevated bio-ADM levels in patients with septic shock, in accordance with earlier reports [13,14]. Our study also confirmed the marked association between bio-ADM level at admission and short-term mortality as well as the prognostic cutoff value of 70 pg/ml, previously described by Marino et al. [14] and Caironi et al. [13] in both sepsis and septic shock (including the most recent definition [17]). Our study showed moderate prognostic value of bio-ADM at admission using AUC but marked prognostic value using Cox proportional hazards model adjusted for various parameters. Moreover, our study showed that prognostic value of bio-ADM at ICU admission exerts additive value (positive changes in chisquare) to various ICU severity scores. We described also the association between a bio-ADM ≤ 70 pg/ml on day 2 and very low 28-day mortality, even in patients with initial high bio-ADM levels. The association of low bio-ADM by day 2 with full restoration of organ function at day 7 has been shown as well.
Concerning organ dysfunction, we found a relationship between circulating bio-ADM at ICU admission and the subsequent need for cardiovascular and/or renal support. In our studied patients, high circulating bio-ADM-known to have vasodilatory actions-might account for the deterioration of vascular tone and blood pressure, as previously described [13,14]. In the present study, patients with high bio-ADM levels on ICU admission were more likely to need vasopressors and/or inotropes either at admission or in the following days. Moreover, they had a higher total fluid balance and higher incidence of RRT during their ICU stay. The ADM-induced vascular dysfunction may have contributed to this condition, although some data suggest that high bio-AM levels might also be protective to the kidney [19,20]. Further studies are needed to elucidate the . Only a small number (n = 16, 2.7%; 28-day survival rate 68.8%) of patients who presented with a low bio-ADM concentration upon admission had higher bio-ADM level on day 2 (low-high (LH) group), which is why this group is not represented in the figure exact role of bio-ADM in renal function. Of interest, the relationship between circulating bio-ADM levels and extent of organ dysfunction, present during ICU admission, was also true during the recovery phase. Indeed, bio-ADM levels decreased before the improvement of total SOFA score in our investigation. Patients with high bio-ADM levels at ICU admission who showed a decline towards normal bio-ADM values at day 2 were more likely to recover vascular function and vasopressor need by day 7. By contrast, the drop in bio-ADM from ICU admission to day 2 was associated with only limited improvement in renal function or no improvement in lung function at day 7. These observations also warrant further exploration.
Circulating bio-ADM levels were lower in AdrenOSS-1 than in the previously described ALBIOS cohort [13]. Indeed, in ALBIOS, septic patients were more severe, as suggested by greater prevalence of mechanical ventilation, length of stay, and short-term mortality. Likewise, the prevalence of septic shock was greater in ALBIOS than in AdrenOSS-1 (Additional file 12: Table S4). Of note, different definitions of septic shock in the two studies may have influenced study assessments.
Limitations included that in the present population only patients with sepsis and septic shock were studied, and results cannot be directly translated to a general ICU population. Future studies should focus on extrapolation of our results to patients with hemodynamic instability related to other disease, because as study has already been performed for cardiogenic shock [21]. Furthermore, our data suggest that ADM may be associated with myocardial function (e.g., patients with high ADM also had significantly higher circulating natriuretic Fig. 3 Association between biologically active adrenomedullin levels upon admission and (a) length of total organ support over the first 7 days (p < 0.0001), (b) length of vasopressor/inotropic support over the first 7 days (p < 0.0001), (c) overall need for vasopressor support (p < 0.0001), and (d) total fluid balance over the initial 48 h (p = 0.0001) peptide levels). However, data on cardiac function (e.g., cardiac output or left ventricular ejection fraction) were available in only few studied patients. Finally, we used the cut point of 70 pg/ml of circulating Bio-ADM for validation of the previously published cut point, even though the optimal Youden cut points in AdrenOSS-1 showed that 70 pg/ml with respect to a technical optimality criterion is not optimal.
Strong points of the study are the fact that it was a prospective international multicenter study with a large number of patients, with a focus on mortality and organ dysfunction. However, as is true of any observational study, only associations can be described, and causeand-effect relationships cannot be deducted.

Conclusions
In this large prospective international cohort of critically ill patients admitted to the ICU with sepsis or septic shock, we confirmed the strict relationship between high levels of bio-ADM at ICU admission and organ dysfunction and mortality. We demonstrated that early decrease towards the normal values of circulating bio-ADM in the first days after ICU admission was associated with improvement of cardiovascular and renal function and was associated with very low 28-day mortality. Fig. 4 The absolute Sequential Organ Failure Assessment (SOFA) scores at (a) admission, (b) day 2, and (c) day 7 for groups high-high (HH; i.e., above 70 pg/ml at baseline and day 2), high-low (HL), and low-low (LL), excluding patients who died within 7 days. p Value for differences between HH vs. LL: p < 0.0001 for all days; p value for HH vs. HL: p < 0.0001 for all days; p Values for HL vs. LL: p < 0.0001, 0.6016, and 0.9969 for days 1, 3, and 7, respectively. Of note, the number of patients is less at day 2 than at day 7 because there were more values missing at day 2 owing to the fact that discharged patients (mostly at day 7) were given a SOFA score of 0. Furthermore, only a small number (n = 16, 2.7%) of patients who presented with a low bio-ADM concentration upon admission had a higher bio-ADM level on day 2 (low-high [LH] group), which is why this group is not represented in the figure. Median (IQR) SOFA scores for the LH group were 7.5 (6.0-9.8), 9.0 (4.0-11.2), and 4.0 (0.0-6.5) for admission, day 2, and day 7, respectively Delabranche, Alexandra Boivin, Raphaël Clere-Jehl,