A Review of Adrenomedullin in Pediatric Patients: A Useful Biomarker

Adrenomedullin has several properties. It acts as a potent vasodilator, has natriuretic effects, and reduces endothelial permeability. It also plays a role in initiating the early hyperdynamic phase of sepsis. Since its discovery, many articles have been published studying the uses and benefits of this biomarker. The aim of this review is to determine the usefulness of adrenomedullin in pediatric patients. Relevant studies covering adrenomedullin in pediatrics (<18 years) and published up until August 2021 were identified through a search of MEDLINE, PubMed, Embase, Web of Science, Scopus, and Cochrane. Seventy studies were included in the present review, most of them with a low level of evidence (IV to VI). Research on adrenomedullin has primarily been related to infection and the cardiovascular field. The performance of adrenomedullin to quantify infection in children seems satisfactory, especially in sepsis. In congenital heart disease, this biomarker seems to be a useful indicator before, during, and after cardiopulmonary bypass. Adrenomedullin seems to be useful in the pediatric population for a large variety of pathologies, especially regarding infection and cardiovascular conditions. However, it should be used in combination with other biomarkers and clinical or analytical variables, rather than as a single tool.


Introduction
Adrenomedullin (ADM) is a new, readily measurable circulating biomarker that was first detected in pheochromocytoma cells [1]. The gene locus, part of the calcitonin gene family, is located on chromosome 11. It is a peptide composed of 52 amino acids. After its release, ADM is rapidly eliminated from the serum (half-life of just 22 min); thus, serum levels are difficult to measure [2]. Its precursor molecules are pre-pro-adrenomedullin, which consists of 185 amino acids, and pro-adrenomedullin, which contains 164 amino acids and remains stable for longer periods [3]. In 2004, Struck et al. [4] discovered a more stable mid-regional portion of 52 amino acids, called mid-regional pro-adrenomedullin (MR-proADM). It is co-synthesised with ADM, directly reflects the levels of degradation of active ADM, and has the advantage of a longer half-life and lack of protein binding, which makes it easier to analyse and more suitable for daily clinical practice [5,6]. Recently, assays have become available to measure MR-proADM.
ADM is mainly released from endothelial tissues such as the adrenal medulla, lungs, kidneys, gastrointestinal organs, smooth muscle, and heart [7]. Its production and secretion are stimulated by inflammatory processes, and it has been shown in recent years that cytokines and other inflammatory agents such as LPS, tumour necrosis factor, interleukin-1, and bacterial endotoxins induce its expression.
ADM has several interesting properties [8,9]. Released from endothelial cells, it acts as a potent vasodilator by increasing the synthesis of nitric oxide. It also has natriuretic effects and reduces endothelial permeability [10]. Clinicians have also observed that it plays a key role in initiating the early hyperdynamic response in sepsis and septic shock [11]. The fact that the expression of ADM is stimulated by endotoxins and bacterial cytokines, and that it is bound by the complement control protein Factor H, leads us to assume that ADM has an antibacterial effect and a role as a downregulator of pro-inflammatory cytokines [12].
Since its discovery, many articles in different fields of medicine have been published studying the potential uses and benefits of this biomarker in adults, children, and neonates. To date, no comprehensive review has been undertaken that compiles all of these studied uses in children. We hypothesize that this biomarker can be incorporated into clinical practice for the diagnosis and severity assessment of multiple pathologies.
The aim of this review is to determine the usefulness of ADM in pediatric patients in different situations.

Methods
To identify relevant studies featuring patients less than 18 years of age, a search was performed using MEDLINE (PubMed), Embase, Web of Science, Scopus, and Cochrane for articles published up until August 2021. This was performed with the search words: (adrenomedullin (title/abstract)) AND children (title/abstract) OR pediatric (title/abstract) OR infant (title/abstract). Reviews, opinion papers, genetics studies, and articles focusing on neonates were excluded. The full texts of the remaining studies, full-text articles assessed for eligibility, were screened by two independent reviewers. Disagreements were resolved through consensus or recourse to a third reviewer. STROBE checklists were used to measure the strength of the articles. The content of this protocol follows Preferred Reporting Items for Systematic reviews and Meta-analysis (PRISMA) recommendations (Please see attached the Supplementary File S1).
Data from reports were extracted by one reviewer. The primary measures considered were the AUC for predicting the outcome, along with the optimal cut-off and the main diagnostic validity parameters (sensitivity, specificity, and positive/negative predictive values). When this information was missing, differences between the means were taken into account. Risk of bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool.
Since the study was designed as a systematic review of the literature, permission from an ethical committee was not deemed to be necessary.

Results
Articles published up until August 2021 involving patients under 18 years old and ADM were reviewed. A total of 126 articles were found. A total of 70 studies were ultimately selected (Tables 1-7). Most of these studies had a low level of evidence (IV to VI), being case-control or cohort studies. The rest of the studies were rejected for the following reasons: eleven were reviews, seven were articles on adults, two were experimental studies, four were articles on genetics, two were meeting communications, two were opinion papers, one was a preliminary article, ten were articles on neonates, three were articles in a language other than Spanish or English, and fourteen were articles that that did not have a sufficiently high-quality methodology ( Figure 1). The summarized QUADAS-2 assessment of the studies is shown in Figure 2. Bias may be present especially regarding patient selection (more than 50% of the studies) and interpretation of the index test (more than 60% of the studies). MR-proADM appears to improve diagnostic accuracy for detecting patients with higher risk of mortality scores and more than one organ failure. PCT appears to be superior to MR-proADM in diagnosing sepsis; MR-proADM in the early stage of sepsis could be a useful tool for the stratification of sepsis and the prediction of morbidity.

MR-proADM
To explore the relationship between the need for respiratory support and MR-proADM.
-Children who needed nasal cPAP or MV presented higher MR-proADM levels than the group that required just high-flow therapy or no extra support (p < 0.05).
MR-proADM could be a potential biomarker for the severity of acute bronchiolitis.  Troponin-I Preoperative ADM levels in the group with little or no evidence of myocardial injury after the surgery were significantly greater than the groups with either moderate or severe injury.

Girona
Higher preoperative ADM levels are associated with lower levels of myocardial injury (as assessed by troponin-I release) during surgery for congenital heart defects. - The ADM levels before CPB were slightly higher than normal, and increases of 323% were observed on day 1 after CPB; levels returned to baseline on day 3. ADM was correlated to insulin variations and was an independent factor associated with lower insulin concentrations in the multiple regression model.
ADM may be a predictor of low insulin concentration after CPB. The VIS score at 2 h post-CPB was identified as an independent early predictor of LCOS. This predictive value was not significantly increased when associated with cardiac biomarkers for LCOS.

Cardiac catheterisation
Plasma AM-LI concentrations in the C-PH group and the P-PH group were significantly higher than in the no PH group.
There is a significant decrease in the plasma AM-LI concentration between the PA and PV.
There is a significant correlation between plasma AM-LI in the PA and pulmonary artery pressure.
ADM may be involved in the cardiovascular regulation or homeostasis of pulmonary circulation in pulmonary hypertension. -Plasma ADM levels were significantly higher in patients with CHD than in the control group (p < 0.05) and increased significantly as PH severity increased from non-PH to mild to moderate-severe (p < 0.05, p < 0.01).
On the 7th day after CPB, plasma ADM levels in the PH group were significantly decreased in comparison with those before the operation. Plasma ADM levels in CHD were positively correlated with PASP.
ADM may play an important role in the development of PH in patients with CHD. ADM may be involved in the defence mechanism against further increases in pulmonary arterial pressure. -As pulmonary hypertension increases in severity, the plasma levels of ADM increase. There is a positive correlation between PSAP and plasma ADM levels.
Plasma levels of ADM in each group after CPB is lower than that of each group before the operation.
Measuring the levels of ADM may be a reliable method to monitor changes in pulmonary pressure and the worsening of pulmonary hypertension.

Clinical criteria
Plasma ADM concentrations during relapse were significantly lower than in remission and controls, and urine ADM levels were significantly higher in relapse than in remission and controls.
The important changes in plasma and urine ADM levels in relapse suggested that these changes may be compensatory when the body experiences severe proteinuria. Cystoureterography DMSA Plasma ADM levels were not significantly higher in the control group than in the rest of the groups (p = 0.162). Urine ADM levels were higher in the control group than in the three study groups (p < 0.005).
Urine ADM can be a prognostic factor in the long-term follow-up of paediatric cases with VUR.

S-cone ERG protocol
A statistically significant finding was that the ADM level in the diabetes group was lower than in the control group (p < 0.05).
The changes observed in the ADM levels support its possible involvement in the microvascular complications of diabetes. Different-than-expected study results as regards ADM concentration indicate that further studies are needed.

Critically Ill Children and Septic Patients
The available tools to determine the prognosis of critically ill children have mostly been scales developed to estimate the mortality risk based on clinical signs and routine analyses. These include the Pediatric RIsk of Mortality III score (PRISM III) [82], Pediatric Index of Mortality 2 (PIM 2), Pediatric Multiple Organ Dysfunction Score (PMODS) [83], Sepsis-Related Organ Failure Assessment (SOFA) [84], and PEdiatric Logistic Organ Dysfunction-2 (PELOD-2) [85]. These scores are useful to predict the evolution of a wide group of patients within the first 24 h, but their utility in individual patients, especially at the time of admission, seems to be limited. ADM has been recently proposed as a useful biomarker for evaluating disease severity and risk of death.
In a heterogeneous cohort of 254 critically ill pediatric patients recruited from two pediatric intensive care units (PICUs), high levels of MR-proADM, carboxy-terminal pro-endothelin-1 (CT-proET-1) and procalcitonin (PCT) were associated with increased risk of mortality scores as well as an increased risk/number of organ failures. The single marker with the greatest area under curve (AUC) was MR-proADM. Combining MR-proADM with each of the other markers slightly improved diagnostic efficiency. MR-pro-ADM demonstrated an excellent accuracy when detecting patients with higher risk of mortality scores and more than one organ failure, with an AUC of 0.866 (95% CI 0.810-0.821) and 0.922 (95% CI 0.887-0.957), respectively (Table 1). Regarding the sub-cohort of septic patients (n = 40), the AUCs were still excellent (0.869 and 0.901), with similar cut-off points: 0.80 nmol/L (Se 88%, Sp 70%, PPV 23%, NPV 95%) for the prediction of high risk mortality scores and 0.81 nmol/L (Se 90%, Sp 74%, PPV 50%, NPV 96%) for the prediction of more than one organ failure [13].
In 2014, Jordan et al. investigated the prognostic usefulness of MR-proADM in 95 pediatric patients with sepsis admitted to the PICU, finding that MR-proADM levels were significantly higher in those patients requiring mechanical ventilation and vasoactive drugs, although no differences were seen in patients with renal failure. A positive correlation between PRISM III and MR-proADM levels at admission was found (r = 0.447, p < 0.001). MR-proADM was also higher in cases of in-hospital mortality (the sample had a mortality of 10.5%), with an AUC of 0.77, slightly lower than the AUC of PRISM III (0.78), and an optimal cut-off point of 2.2 nmol/L. It showed better positive predictive values than PCT and C-reactive protein (CRP) (31% vs. 21.6% and 15.8%, respectively), with similar negative predictive values. In the multivariate study, PRISM III and MR-proADM were identified as independent predictors of in-hospital mortality; thus, their combined use was proposed [14].
Later on, the same group evaluated the diagnostic, prognostic, and stratification potential of a single MR-proADM measurement at the onset of febrile syndrome (Solé-Ribalta et al., 2020). Mean MR-proADM values showed an increasing trend in the transition between non-infected patients, patients with a localized infection, and septic patients, but the difference between non-septic groups and septic groups was not significant. MR-proADM levels increased significantly in patients with severe sepsis (p = 0.048) and septic shock (p = 0.004) compared with sepsis, with an AUC for the prediction of severe sepsis of 0.729 (p = 0.013) and an optimum cut-off of 1.37 nmol/L (Se 79.4%, Sp 50%). Its correlation with mortality risk scores was confirmed. MR-proADM levels were also significantly increased in patients requiring vasoactive drugs; however, unlike in the previous study, they were associated with renal dysfunction but not with the need for mechanical ventilation. It was concluded that PCT appears to be superior to MR-proADM in the diagnosis of sepsis, and that the determination of plasma MR-proADM levels in the early stages of sepsis could be a useful tool for its stratification and predicting morbidity [16].
The results of this last study contrast with the results of Lan J. et al., who also explored the role of MR-proADM in the early diagnosis of childhood sepsis using a sample extracted during the first hour of admission. The authors saw that the diagnostic ability of MR-proADM in children with sepsis was better than that of PCT (AUC 0.869 and 0.757, respectively). Its combined detection effect was better than either single test. MR-proADM and PCT levels also gradually increased in relation to the severity of sepsis and in relation to the systemic inflammatory response syndrome (SIRS) group and control group [15].

Acute Appendicitis
Acute abdominal pain in children is a common reason for visits to the emergency department. Usually, it is a self-limiting process, but in a small percentage of cases, it will require surgery (i.e., acute appendicitis, AA). Diagnosing appendicitis in children is problematic, because most cases show signs and symptoms that mimic other, self-limiting causes of abdominal pain. A delay in the diagnosis and treatment of a case of appendicitis leads to an increase in appendix rupture rates, post-surgical morbidity, mortality, and length of hospital stay, but minimizing the number of negative appendectomies is also important. A large number of inflammatory markers have been used for the early diagnosis of appendicitis. None, in any combination, have shown a satisfactory predictive value for the early diagnosis of appendicitis in the pediatric population. For this reason, some authors have turned their attention to ADM.
According to articles published on this topic, MR-proADM concentrations are higher in children with acute appendicitis than in children with non-specific abdominal pain [17,18]. Greater differences in the levels of this biomarker were seen in complicated cases than in uncomplicated cases of appendicitis, in the first study being not significant (p = 0.159), probably due to the smaller number of patients, and being significant in the other (p = 0.005). The performance of MR-proADM alone, although statistically significant, was not optimal, with an AUC of 0.75 in the first study and 0.66 in the second, similar to that of CRP (0.72 and 0.70, respectively) and lower than that for leukocyte count (0.88 and 0.84), neutrophil count (0.86 and 0.84), and PAS score (0.87 and 0.76). An MR-proADM level of <0.35 nmol/L, in combination with a low CRP level, seemed useful for identifying children with a low risk of AA, with a sensitivity of >95% and a negative predictive value of >90%. Thus, it seems that ADM alone is not enough for the early diagnosis of AA, but in combination with CRP, it can help to select children with a low risk of AA, and it may contribute to assessing the severity of AA.

Pneumonia
Community-acquired pneumonia (CAP) is a major cause of morbidity and mortality worldwide, especially among children under 5. Basing their opinion only on the assessment of clinical signs and symptoms, clinicians may underestimate or overestimate the severity of the patient's illness. This can cause unnecessary hospitalization of a patient or trying to treat a patient in an outpatient setting when they require hospitalization, which may cause complications. Due to this, there is a growing interest in searching for biomarkers to improve the risk/severity stratification, alone or in combination with clinical prognostic scores.
Several studies investigating the relationship between pro-ADM levels and CAP patients' prognosis have been published. Patients presenting complications had significantly higher levels of pro-ADM [19], especially as regards pleural effusion, bacterial pneumonia, empyema, and the need for interventional procedures. Other biomarkers studied, such as copeptin (CoPEP) or interleukin-β1 (IL-β1), showed no added value [20] or inferior results compared to ADM [22]. In this last study, the combination of a Clinical Respiratory Score over 6 points and pro-ADM values over 1.75 nmol/L showed the most significant results (OR: 15.38, 95% CI 1.35-166.66, p = 0.027). Despite the homogeneity of the studies, diverse cut-off points have been described: 0.16 nmol/L and 1.75 nmol/L for CAP presenting with complications.
The diagnostic value of ADM in pediatric CAP has only been studied by Esposito et al., who found that MR-proADM blood levels are unable to differentiate bacterial from viral diseases. In this same study, the discriminatory power of MR-proADM to identify complications was found to be insufficient, and no advantages were seen with the use of soluble triggering receptor expressed on myeloid cells-1 (sTREM-1), mid-regional pro-atrial natriuretic peptide (MR-proANP), and MR-proADM, while PCT remained the most useful biomarker for both cases [21].

Pyelonephritis
Urinary tract infection (UTI) is one of the most common infectious pathologies in infancy. Although most cases have an excellent prognosis, when there is renal parenchymal involvement, this leads to an increased associated morbidity and patients may develop proteinuria, high blood pressure, and chronic renal failure in the future. Renal cortical scintigraphy with 99mTc dimercaptosuccinic acid (DMSA scan) is the gold standard for the diagnosis of acute pyelonephritis (APN) and renal scarring (RS). However, it is an invasive test that exposes the patient to radiation and it is not available in all centers. Conversely, there are biomarkers of parenchymal involvement such as CRP or PCT, but none of them are sensitive and specific enough to replace the DMSA scan. For this reason, new molecules that can aid in an early detection of those patients at risk of kidney damage continue to be searched for.
Urine ADM has been studied as a possible solution [23][24][25]. Urine ADM had been observed to be significantly higher in patients with UTI than in healthy controls, and it seems that successful antibiotic treatment of a UTI normalizes ADM levels. In addition, there is a tendency for ADM values to be higher in the urine of patients with ANP than children with acute cystitis. There are conflicting results in the literature about the correlation between white blood cell counts (WBC), CRP, erythrocyte sedimentation rate (ESR), and urine ADM. Sharifian M. et al. proposed a cut-off value of urine ADM for the diagnosis of APN of 100 pg/dL (Se 67.7%, Sp 70%, PPV 70%, NPV 67.7%).
Regarding plasma levels, in one study proADM levels had good predictive value for the diagnosis of APN among patients with UTI (AUC = 0.830, cut-off value for proADM of 63.86 pg/mL [26]. A later study by Peñalver R. et al. could not find a relationship between APN and plasma ADM; however, they did find that plasma MR-proADM had excellent prognostic capacity to predict RS development in patients with APN.

Febrile Neutropenia
Infection is a leading cause of death in children with malignancies. Susceptibility to infection is increased in such patients because they may have to receive intensive chemotherapy, and consequently, they often develop immunosuppression and neutropoenia. The blood culture test, which is considered the gold standard in diagnosing sepsis, is time consuming and associated with a high risk of false negatives (and the blood culture is positive in only up to 20-30% of cases). WBC cannot be applied. The biomarkers that can help uncover the etiological agent can be confounding because of the harmful effects of chemo or radiation therapy. Therefore, a risk score system is commonly used to evaluate and treat patients. However, there is still a need for new, specific biomarkers.
Demirkaya et al. [28] measured serum ADM levels and could not demonstrate a correlation between these levels and the severity of febrile neutropenia. They attributed this phenomenon to its rapid clearance from the serum. Therefore, instead of ADM, they proposed measuring pro-ADM with its longer clearance time. In a study conducted by Fawsi M.M. et al. [31], with severely neutropenic young patients with hematological malignancies, MR-proADM levels were significantly different between the bacteremia/sepsis and pyrexia of unknown origin groups. The area under the receiver operating characteristic curve of MR-proADM had an excellent discriminatory power (AUC = 0.939). The cut-off value defined was 2.4 nmol/L (Se 91.6%, Sp 85.1%, PPV 83.3%, and NPV 92.4%). This study adds to what other recent publications are saying: MR-proADM is a promising early biomarker for sepsis in severely neutropenic young patients with hematological malignancies. Regarding risk prediction, ADM levels at admission seem to be useful in identifying high-risk patients with solid tumors [29], and they have been correlated with length of stay (LOS) in hospital [30].

Congenital Heart Disease
Studies have demonstrated that ADM has various physiological effects on the cardiovascular system, including vasodilatation, diuresis, natriuresis, inhibition of aldosterone secretion, and increased cardiac output. These findings suggest that plasma ADM may participate in the regulation of circulatory homeostasis and in the pathophysiology of cardiovascular disease. Regarding the behavior of this biomarker in congenital heart disease, it seems that ADM levels may play beneficial roles in reducing increased pulmonary arterial resistance or alleviating hypoxemia in these patients, since its levels are higher in patients with congenital heart disease (CHD) and pulmonary hypertension (PH), and it is positively correlated with pulmonary arterial resistance (Rp). ADM levels are also higher in cyanotic congenital heart disease and are negatively correlated with systemic arterial oxygen saturation (SA sat ) and mixed venous oxygen saturation. In addition, its levels were lower in the pulmonary vein than in the pulmonary artery, suggesting that it is extracted in pulmonary circulation [36][37][38].

Cardiac Surgery
All organs are harmed during cardiopulmonary bypass, and the pulmonary endothelium is one of the most severely damaged, causing high pulmonary vascular resistance, among other consequences. ADM, an intrinsic vasodilator thought to act as a circulating hormone that regulates vascular tone, has been analysed in several studies involving children undergoing open-heart surgery. Plasma ADM levels significantly increase after cardiopulmonary bypass (CPB), but patients with high pulmonary arterial flow showed lower levels than other patients, both pre-and post-CPB. One explanation could be previous endothelial damage in these patients, also supported by the correlation between post-CPB ADM levels and pre-CPB mean pulmonary arterial pressure (mPAP) [39].
In addition, the preservation of the myocardial tissue remains a challenge, and the distribution of cardioplegia solution is altered due to partial vasoconstriction. Thus, it was thought by Szekely et al. that the presence of vasodilators such as ADM could improve this distribution. Their results showed that preoperative ADM levels in patients with cardiac troponin I (cTn-I) levels <10 ng/mL after the surgery were significantly greater than those with either moderate or severe injury; thus, their hypothesis was sustained [40].
The CPB procedure in itself implies a complex fluid homeostasis with the involvement of many hormones, such as vasopressin, atrial and brain natriuretic peptides, and aldosterone. ADM kinetics during CPB has also been studied. ADM levels increased gradually, peaking 60 min after CPB, and then decreased 24 h after the operation, in tandem with changes in urine sodium and vasopressin. Besides these two parameters, ADM was also correlated with urine volume, plasma osmolarity, and plasma brain natriuretic peptide (BNP). Changes in plasma ADM levels seem to correspond to fluid and electrolyte balance, and ADM may play a role in fluid homeostasis, in cooperation with other hormones, during CPB [41].
CPB affects the cerebrovascular autoregulation system as well, mainly due to the hypoperfusion-reperfusion sequence, embolization, and/or thermal injury. Florio et al. found that at the end of CPB and surgery, ADM was significantly lower in children in whom brain damage was shown during follow-up for cardiovascular surgery; these children also had increased cerebrovascular resistance. They proposed a cut-off value of 17.4 ng/L for ADM at the end of CBP for predicting brain damage [42].
ADM has been proposed as a biomarker for low cardiac output syndrome (LCOS) after CPB in children, which would be of great interest because this is still a major perioperative complication. ADM levels in patients that developed LCOS were significantly lower than those who did not, reaching a dip at the end of the surgical procedure. One multivariable analysis demonstrated that ADM levels and cooling were independent variables for predicting LCOS, and a cut-off point of 27 pg/L for ADM at the end of the surgery was proposed for this [43]. In another study, cTn-I levels (>14 ng/mL; OR 4.05, 95% CI 1.29-12.64, p = 0.016) at 2 h following corrective heart surgery under CPB and MR-proADM levels (>1.5 nmol/L; OR 15.54, 95% CI 4.41-54.71, p < 0.001) at 24 h post-operation were independent predictors of LCOS [45]. This, however, differs from two previous papers [41,43] that did not find an initial decrease in ADM levels during and after CPB in children with LCOS. One year later, this same author published that a vasoactive inotropic score (VIS) of >15.5 at 2 h post-CPB, adjusted for age and CPB timepoints, showed a high specificity (92.87%) and negative predictive value (75.59%) for diagnosing LCOS at 48 h post-CPB, although the predictive power for LCOS did not increase when VIS was combined with cTn-I >14 ng/mL at 2 h and MR-proADM >1.5 nmol/L at 24 h post-CPB. This could provide important information for physicians so they can engage in early interventions and optimally manage these patients [46].
Regarding the need for intensive support and complications after CPB, MR-proADM and pro-atrial natriuretic peptide (pro-ANP) were associated with the need for MV, with respective cut-off values of 1.  [47]. Pro-ANP and MR-proADM levels before cardiac surgery were also higher in those patients with an increased need for ventilatory and inotropic support after the surgical intervention. The AUC for pro-ANP was greater than that of MR-proADM, but without significant differences for both cases; however, in the multivariate analysis, the only biomarker kept as an independent predictor was pro-ANP [48].

Fontan Procedure
Pulmonary factors, especially pulmonary vascular resistance (PVR), are known to be crucial when deciding on the Fontan procedure. Therefore, the cytokines that influence pulmonary circulation also seem to influence hemodynamics after the Fontan procedure. Immediately after the Fontan procedure, both endothelin-1 (ET-1) and ADM are elevated, but plasma levels of ET-1 are maintained and then significantly increase 6 and 24 h after CPB, unlike ADM. Therefore, the imbalance of these two cytokines may trigger vasoconstriction after the Fontan procedure that may compensate for the relatively low cardiac output [49]. ADM has also been studied during follow-up on Fontan procedure patients. Higher concentrations were found in these patients than in controls, and a negative correlation was also observed between ADM levels and cardiac output [50]. MR-proADM levels may help identify patients at risk for a failed Fontan, and a cut-off of 0.520 nmol/L has been proposed for predicting Fontan failure [51].

Heart Failure and Dilated Cardiomyopathy
Measuring blood biomarkers can facilitate heart failure (HF) management, as they provide objective information on disease severity, prognosis, and treatment response. Although plasma levels of ADM were increased in pediatric and adult patients with CHF irrespective of the cause [52], a larger study focusing on the pediatric population found an unsatisfactory diagnostic power for MR-proADM regarding general patients with HF and the subset of patients with dilated cardiomyopathy (DCM) [53]; it was, however, negatively correlated with poor left ventricle (LV) function. In this study, MR-proANP was found to be the best biomarker discriminating HF in children and adolescents with CHD and CMP, even if well compensated, and its diagnostic performance was comparable to that of NT-proBNP. Conversely, another study including only a small sample of children with DCM observed significantly lower levels of ADM in DCM patients than controls. They hypothesized that this finding could be due to events such as the necrosis and apoptosis of cardiomyocytes in advanced stages of heart failure, but this hypothesis must be examined in additional studies. Despite their different results, they agree on the negative correlation between ADM levels and ejection fraction (EF) and fractional shortening (FS) [54].

Postural Orthostatic Tachycardia Syndrome
Children with postural orthostatic tachycardia syndrome (POTS) experience symptoms of orthostatic intolerance in association with excessive tachycardia. Midodrine hydrochloride, a vasoconstrictor, has been reported to improve symptoms, although the response rate is around 70%. A predictor for its response has not been described, but it is presumed that children with abnormal vascular resistance might have a positive response to the drug. In 2012, Zhang et al. studied the predictive value of MR-proADM for assessing the therapeutic efficacy of midodrine hydrochloride. They found that those responding positively to midodrine hydrochloride had higher plasma levels of MR-proADM than nonresponders, with an excellent discriminatory power (AUC of 0.879), and they proposed a cut-off of 61.5 pg/mL to predict the efficacy of the drug [55]. The same group reported that patients with levels above the defined cut-off had a significantly lower symptom score at the 60-month follow-up and a significantly higher symptom-free survival at the 72-month follow-up [56].

Pulmonary Hypertension
ADM exerts its vasodilatory activity on a number of vascular beds, including the lungs, and stimulates cyclic adenosine monophosphate (cAMP) production in vascular smooth muscle cells, suggesting that it may play an important role in modulating PH. An increased pulmonary vascular resistance has been documented during and after CPB and this is a major concern in children with CHD.
ADM is significantly elevated in patients with PH (primary and related to congenital heart disease) when compared to those without PH, and a significantly increased uptake of plasma ADM in pulmonary circulation has been demonstrated in these patients [58]. Additionally, as the severity of PH increases, the plasma levels of ADM also increase in tandem (the same as ET-1 and contrary to nitric oxide (NO) levels), and a correlation be-tween ADM levels and PAP was observed [61,62]. These findings suggest the involvement of ADM in the pathophysiology of PH. ADM may be involved in the defense mechanism against further increases in pulmonary arterial pressure and may be a reliable method to monitor changes in pulmonary pressure and worsening PH.
With respect to the presurgical stage, Vijay et al. demonstrated that preoperative plasma ADM levels are lower in patients who are at a higher risk for the development of postoperative PH (patients with high pulmonary flow), suggesting that the impaired ability to synthesize ADM in the pulmonary circulation may contribute to the risk of developing PH in pediatric patients undergoing CPB [59].
As for long-term follow-up, ADM may be valuable for evaluating pulmonary hemodynamics after long-term treatment with PGI2 in primary PH [60].

Nephro-Urological Disease
ADM is a potent vasodilator with proven antimitogenic and antiproliferative effects in renal mesangial cells, as well as diuretic and natriuretic actions. Plasma and urine ADM levels are known to deviate from normal levels in many renal diseases.
Balat et al. studied plasma and urine levels of ADM in minimal change nephrotic syndrome (MCNS) [64] and found that ADM levels were significantly lowered in plasma and were elevated in urine in these patients during relapse. This is possibly due to ADM loss through the urine secondary to severe proteinuria and because the kidney may be one of the major sites for ADM synthesis. In addition, NO and ADM, with their vasodilatory effects, were considered of interest in children with detrusor instability (DI) [66]. It was found that a decreased NO production most likely has a role in the pathophysiology of DI, although increased ADM appears to be compensatory, and the functional significance of ADM and NO in bladder smooth muscle remains to be determined by further, more in-depth studies. The same author studied the role of NO and ADM in Bartter syndrome [63], in which patients are known to have a lower vascular reactivity. It was found that urinary nitrite and ADM excretion was lower in Bartter syndrome patients, suggesting a potential role of these two molecules in the reduced vascular response seen in the disease. Finally, Balat and his group investigated the role of ADM and NO in primary nocturnal enuresis [65], finding that bed-wetting children had plasma levels that were significantly lower than the controls, suggesting a deficient synthesis in these children.
In 2005, Kalman et al. studied ADM in children with urinary tract infections with renal parenchymal scars (RPS) and vesicoureteral reflux (VUR) [67]. Plasma and urine ADM levels were higher, although not significantly, in the control group. The authors justify this finding with the fact that prophylactic antibiotics may lower ADM secretion. However, the low urinary ADM levels of patients with renal parenchymal scars and preserved renal function might be predictive of renal injury, and the lower ADM levels in patients with a higher degree of VUR may indicate that ADM has a role in the ureteral antireflux mechanism. Thus, they conclude that ADM can be a prognostic factor in the long-term follow-up of cases with these diseases.

Endocrine Pathology
Obese children are at an increased risk for abnormal cardiac structure and function and other metabolic risks. Studies on adipokines and other biomarkers of obesity have become important in obesity research, and ADM was also defined as a new member of the adipokine family. ADM secreted by adipocytes, through its vasodilator and antioxidant actions, might be protective against metabolic-syndrome-associated cardiovascular complications.
Plasma MR-proADM levels are significantly higher in obese than in normal-weight adolescents, and its levels are correlated with BMI z-score, fat mass, circulating insulin, HOMA-IR, total cholesterol, and LDL cholesterol, suggesting its important involvement in obese patients [68]. A positive correlation between ADM levels and left ventricle mass index (LVMI) have also been demonstrated (LVMI OR 1.14, 95% Cl 1.08-1.13, p = 0.0001). In this same article, a cut-off value of ADM at 52 pg/mL is proposed to differentiate obese children with and without left ventricular hypertrophy, yielding a sensitivity of 94.32% and specificity of 92.45%. Thus, measuring plasma ADM levels in obese children may help to identify those at high risk of developing LV hypertrophy and dysfunction [69].
Due to its vascular effects, including endothelial vasodilatation, antioxidative stress, stimulation of endothelial nitric oxide production, and antiproliferation of vascular smooth muscle cells and adventitial fibroblasts, it has been speculated that ADM could have a role in microvascular complications in patients with type 1 diabetes mellitus. However, the two articles available are contradictory as regards whether its levels would be increased or decreased.

Rheumatic Diseases
The main source of circulating ADM is now thought to be the vasculature. As both vascular endothelial cells and smooth muscle cells prominently express ADM and its receptor, ADM may control vascular functions and may have a role in rheumatic diseases such as Kawasaki disease (KD), Henoch-Schönlein purpura (HSP), acute rheumatic fever (ARF), and Familial Mediterranean Fever (FMF).
The formation of coronary artery aneurysms is a major complication of KD, and early detection and intensive treatment for it is important. ADM levels were markedly elevated before treatment, especially in patients with coronary artery dilation. They substantially decreased after treatment. Thus, it may be a useful biomarker to monitor KD patients during the acute phase and may help to diagnose coronary artery involvement [72].
ADM may play a role in endothelial injury in HSP since there are higher plasma and urine levels of ADM in patients in the acute phase versus in remission and controls. However, whether this perpetuates, or protects against, further vascular injury is not clear [73].
ARF has the characteristics of an autoimmune disease, triggered by cross-reactive antigens shared by group A streptococci and involving a variety of tissues, such as the heart and endothelium. Plasma and urine ADM levels were significantly higher in children with ARF, irrespective of whether they were in the acute or convalescent phases. ADM may play a role in the immunoinflammatory process of ARF. By contrast, these increased levels may also be the result of inflammatory injury in ARF [54].
FMF is characterized by recurrent attacks of polyserositis. As ADM is synthesized in the endothelium and mediates many functions within the immune system, it has been considered an interesting FMF research target. Plasma and urine levels were significantly higher in patients than in controls; thus, ADM may have a role in the immuno-inflammatory process of FMF, although whether it acts to sustain or protect against further inflammatory injury is not clear [74]. Renal amyloidosis, which is the most important FMF complication, also determines its prognosis. Although information is still conflicting, M694V homozygosity is a risk factor for amyloidosis; thus, the relationship between this genotype and ADM has been studied. Although these results indicated higher ADM levels in patients with a homozygous M694V mutation, further studies are needed to relate ADM levels and renal amyloidosis (62). Finally, in order to investigate the subclinical inflammation present between FMF attacks, Polat et al. conducted a clinical trial with FMF patients at different colchicine dosages. No changes in ADM were demonstrated, suggesting the continuation of the clinical inflammatory process in these patients [76].

Discussion
From this literature review, we can see that research on ADM has focused on infection and cardiovascular diseases. Most of the studies had a low level of evidence (IV to VI), being case-control or cohort studies. Measurements of ADM are quite heterogeneous, in terms of the ADM form (adrenomedullin, pro-adrenomedullin, mid-regional pro-adrenomedullin, and even sub-forms), dosing techniques, and samples chosen for measurement (plasma, serum, urine, tissues, and CSF).
In general, the performance of ADM as a biomarker for pediatric infection seems satisfactory. In pediatric sepsis, it seems to be useful for stratification, prediction of organ dysfunction, and mortality risk prediction, with consistent cut-off points [14][15][16]. These results are similar to those found in the adult population where several studies show that MR-proADM can be helpful in the individual assessment of risk in septic patients admitted to intensive care units (ICU). In 2006, Christ-Crain et al. [86] concluded, after evaluating 53 patients admitted to an ICU for sepsis, that MR-proADM levels at ICU admission are able to predict the patient's condition with an accuracy similar to that of APACHE II and SAPS II. Suberviola et al. [87] showed that MR-proADM is a better tool for predicting in-hospital death than C-reactive protein or procalcitonin in patients admitted to the ICU. Its capacity to identify invasive bacterial infections in well-appearing febrile infants is poor, but this improves in combination with other biomarkers such as PCT [33]. Regarding pneumonia in the adult population, in the study by Cavalli R. et al. [88] from 2014, MR-proADM levels were correlated with mortality; Pereira J.M. et al. [89], remarked that patients with severe CAP who have decreased levels of MR-proADM after 48 h of antibiotic treatment have a better vital prognosis, as well as the study by Akpinar S. et al. [90], who observed that the determination of MR-proADM alone does not correlate with mortality or disease severity to produce mortality but that the union of SOFA, APACHE II and MR-proADM may be good prognostic markers in CAP patients. In pneumonia, it seems to identify patients with complications, although with differing cut-off points [19,20,22]. In cancer patients with neutropoenia, it does not predict a positive blood culture, but it appears to identify high-risk patients and sepsis [28,29,31]. More research is needed in this regard, and it is potentially necessary to differentiate between patients with solid tumors and patients with hematological malignancies. In patients with acute abdomen, its combined use with CRP is proposed to identify low-risk patients [17,18]. Finally, in UTIs, it seems to identify patients with pyelonephritis and patients with pyelonephritis who will develop renal scarring [26,27].
Regarding the behavior of this biomarker in CHD, it seems that ADM levels may play beneficial roles in reducing increased pulmonary arterial resistance or alleviating hypoxemia [37]. Conflicting results exist in the literature regarding the peri-CPB kinetics of ADM. Measuring pre-CPB levels could be useful to predict myocardial tissue damage [40]. During the surgery, ADM levels seem to correspond to fluid and electrolyte balance and may help to predict brain damage [41]. After the surgery, ADM concentrations were useful to predict LCOS and the need for respiratory and inotropic support [43,45]. Vasoconstriction is common after the Fontan procedure; low ADM levels have been found immediately after this surgery, and this has been understood as being protective when there is low cardiac output. Conversely, high levels of this biomarker in the long term may help to identify patients at risk for a failed Fontan [51]. Several studies have analysed the role of adrenomedullin in heart failure in adults, which is linked to cardiac remodeling under stress [91,92]. The higher levels of adrenomedullin were associated with greater volume overload [93], both intravascular and interstitial. In these patients, linked to this situation of overload, the elevated level of adrenomedullin was related to a higher risk of morbidity and mortality [94,95]. In pediatrics, studies regarding heart failure are scarce; unsatisfactory results of ADM were observed in identifying heart failure and dilated cardiomyopathy [96]. Regarding postural orthostatic tachycardia syndrome, ADM could identify those patients who would respond well to vasoconstrictive therapy [55,56].
In PH, ADM may be involved in the defense mechanism against further increases in pulmonary arterial pressure, since levels are elevated in these patients and they negatively correlate with pulmonary arterial pressure [61]. Nevertheless, one article found lower ADM levels in patients at risk of PH (high pulmonary flow patients), and they justify that the impaired ability to synthesize ADM in the pulmonary circulation may contribute to the risk of developing PH in pediatric patients undergoing CPB [59].
Studies on nephro-urologic and rheumatic diseases are scarce and heterogeneous. ADM levels in childhood have been studied in Bartter syndrome, minimal change nephrotic syndrome, primary nocturnal enuresis, detrusor instability, and VUR [63][64][65][66][67]. This peptide seems to have a role in all these pathologies, and in the case of VUR, it seems to identify patients with parenchymal scarring. ADM may have a role in the immuneinflammatory process of HSP, ARF, and FMF, although how they affect inflammatory injury is not clear [73,75,76]. In KD, this biomarker may be useful to monitor patients during its acute phase and may help to determine coronary artery involvement [72].
Our systematic review has some limitations. Most of the studies included had a low level of evidence (IV to VI), being case-control or cohort studies. QUADAS-2 analysis revealed that bias may be present in more than 50% of the studies, especially regarding patient selection and interpretation of the index test. In many cases, the primary measures considered, AUC for predicting the outcome, optimal cut-off and the main diagnostic validity parameters were missing. This fact, along with the heterogeneity in the forms of adrenomedullin measured, made it unfeasible to present a regulated synthesis of results.

Conclusions
ADM seems to be useful in the pediatric population for a large variety of pathologies, but it is especially valuable in infection and cardiovascular conditions. Taking into account all of the above, MR-proADM should be used in combination with other biomarkers and clinical or analytical variables rather than as a single tool. The current clinical data are still limited in some pathologies and have a low level of evidence. More studies are needed to confirm the usefulness of measuring this peptide in some fields.  Institutional Review Board Statement: Ethical review and approval were waived for this study due to it is a systematic review.