Objective Experimental studies have shown that adrenomedullin (ADM) has an important role in circulatory homeostasis. Mid-regional pro-ADM (MR-proADM) is a stable form of ADM. Observational studies found an important association with age, body mass index and kidney function. The aim of this study was to evaluate the prognostic performance of MR-proADM in the general population, controlling for these potential confounders.
Methods 7903 subjects (mean age 49±13 years, 49% male) from the Prevention of REnal and Vascular ENd-stage Disease (PREVEND) cohort with a median follow-up of 10.5 years were enrolled in a prospective cohort study.
Results Mean baseline MR-proADM was 0.39±0.14 nmol/l. In cross-sectional analyses, age, blood pressure, C reactive protein, cystatin-C, N-terminal pro-brain type natriuretic peptide and urinary albumin excretion remained as independent determinants of MR-proADM. In prospective analyses, MR-proADM was associated with the primary endpoint (combined cardiovascular mortality and cardiovascular morbidity), with event rates ranging from 8% in the lowest quintile to 45% in the highest quintile (p for trend <0.001) independent of age, sex, components of the Framingham risk score and other cardiovascular markers. Overall Net Reclassification Improvement against the Framingham risk score was 2.2%, which was non-significant. However, significant modification of the effect of MR-proADM on outcome by age was observed. In subjects aged ≤70 years (N=7475), 8.8% were correctly reclassified in a higher risk category (p=0.017) and 3.4% in a lower risk category (p<0.001). In subjects aged >70 years (N=428) there was no improvement of reclassification (p=0.32).
Conclusion This study gives a detailed overview of the distribution of ADM in a general population and provides evidence that it is a potent and interesting biomarker in predicting cardiovascular events. These results seem especially applicable to younger subjects.
- cardiovascular disease
- heart failure
- renal disease
- metabolic medicine
Statistics from Altmetric.com
Adrenomedullin (ADM) was first discovered in the early 1990s in phaeochromocytoma cells.1 It is a multifunctional 52-amino acid peptide hormone expressed in numerous tissues.2 It is thought to originate primarily in endothelial cells where cellular stress, ischaemia and hypoxia result in increased expression,2 together with nitric oxide and endothelin.3 The physiological function of ADM is still under investigation, but it has been suggested it exerts effects similar to those of brain type natriuretic peptide (BNP) and atrial natriuretic peptide.4 ,5 It is capable of promoting vasorelaxation, natriuresis, diuresis and cardiac output, thus contributing to maintaining circulatory homeostasis.6 ,7 A substantial amount of data also suggests that ADM acts as a protective factor for blood vessels, primarily by counteracting vascular damage and remodelling.3 ,8 These qualities make ADM an interesting candidate novel biomarker for cardiovascular (CV) outcome.
The mid-regional portion of pro-adrenomedullin (MR-proADM) is more stable than ADM and therefore better suited for clinical practice and assessment in stored samples.9 MR-proADM is elevated in women and increases with age.10 It is also elevated in subjects with hypertension and has prognostic value for CV mortality and morbidity in subjects with myocardial infarction,11–13 heart failure,13–15 chronic kidney disease6 as well as subjects with hypertension and increased left ventricular mass.16 MR-proADM was recently shown to be effective in predicting 90-day mortality risk in patients admitted with acute heart failure, with additive prognostic value over BNP alone.17 Epidemiological data on MR-proADM levels in the general population remain scarce, with published data from only one cross-sectional study.10 The present study aims to provide insight into the distribution of MR-proADM and investigate its prognostic performance for CV mortality and morbidity in the general population. Observational studies show an important association with age, body mass index (BMI) and kidney function,10 ,18 but it is unclear whether these associations affect the predictive value of MR-proADM for CV outcome. In addition, the prognostic value and potential additive value of MR-proADM are compared with those of conventional CV risk factors (Framingham risk score, FRS) and adjusted for novel and relevant covariates including N-terminal-proBNP (NT-proBNP), urinary albumin excretion (UAE) and high-sensitivity C reactive protein (hsCRP).
This study was performed in subjects participating in the Prevention of Renal and Vascular End-Stage Disease (PREVEND) study.19 The objective of the PREVEND programme is to investigate prospectively the natural course of increased levels of UAE and assess the value of microalbuminuria as an indicator of increased CV and renal risk in the general population. Details of PREVEND have been described elsewhere.19–21 In summary, from 1997 to 1998, all inhabitants of the city of Groningen, The Netherlands aged 28–75 years (n=85 421) were asked to send in a morning urine sample and complete a short questionnaire on demographics and CV history. Responses were received from 40 856 subjects (47.8%). All subjects with a urinary albumin concentration (UAC) >10 mg/l (n=7786) in their morning urine together with a randomly selected control group with a UAC <10 mg/l (n=3395) were invited to the outpatient clinic. After exclusion of subjects with insulin-dependent diabetes mellitus, pregnant women and men and women unable or unwilling to participate, a total of 8592 subjects completed the screening programme, as shown in figure 1. A total of 7903 individual blood samples taken at baseline were suitable for analysis of MR-proADM levels and eligible for the current analysis. The PREVEND study was approved by the local medical ethics committee and conducted in accordance with Declaration of Helsinki guidelines. All subjects provided written informed consent.
Plasma samples were drawn from all PREVEND participants at baseline and aliquots were stored at −80°C prior to analysis. Detection of MR-proADM was performed using an immunoassay (BRAHMS GmbH/ThermoFisher Scientific, Hennigsdorf, Germany).9 The interassay coefficient of variation was <20% for values >0.12 nmol/l (analytical range 0.08–14.7 nmol/l). NT-proBNP measurements were performed in plasma on an Elecsys 2010 analyser, a commercially available electrochemiluminescent sandwich immunoassay (Elecsys proBNP, Roche Diagnostics, Mannheim, Germany).22 The intra- and interassay coefficients of variation were 1.2–1.5% and 4.4–5.0%, respectively (analytical range 5–35.000 pg/ml). UAE, hsCRP and serum cystatin C were determined by nephelometry (BNII, Dade Behring Diagnostic, Marburg, Germany). Intra- and interassay coefficients of variation were <2.2% and 2.6% for UAE, respectively; <4.4% and 5.7% for hsCRP, respectively; and <4.1% and 3.3% for cystatin C, respectively. Serum creatinine, plasma cholesterol and glucose were determined in one laboratory by Kodak Ektachem dry chemistry (Eastman Kodak, Rochester, New York, USA) using an automated enzymatic method. The intra- and interassay coefficients of variation of serum creatinine were 0.9% and 1.1%, respectively. Serum triglycerides were measured enzymatically. A commercially available assay system was used to assess high-denstiy lipoprotein (HDL) cholesterol (Abbott, Abbott Park, Illinois, USA).
Risk factor definition
Hypertension was defined as systolic blood pressure >140 mm Hg or diastolic blood pressure >90 mm Hg or the use of antihypertensive medication. BMI was calculated as the ratio of weight to height squared (kg/m2). Hypercholesterolaemia was defined as total serum cholesterol of >6.5 mmol/l (251 mg/dl) or the use of lipid-lowering therapy. Diabetes was defined as fasting plasma glucose >7.0 mmol/l (126 mg/dl) or non-fasting plasma glucose >11.1 mmol/l or the use of antidiabetic medication. Estimated glomerular filtration rate (eGFR) was calculated using the simplified Modification of Diet in Renal Disease formula.23 Smoking was categorised as no smoking or smoking (current or stopped <1 year ago). Ten-year risk for CV events according to the FRS was calculated as described by D'Agostino et al24 and divided into three risk categories: low (<10%), intermediate (10–20%) and high (>20%), as recommended by Wilson et al.25
Follow-up for the present investigation is defined as time between baseline urine collection and the date of a first CV event or 1 January 2009. The composite primary endpoint was defined as the combined incidence of CV mortality and CV morbidity after baseline screening. CV morbidity was defined as hospitalisation with a primary discharge diagnosis of documented non-fatal myocardial infarction or myocardial ischaemia, cerebrovascular accident and/or peripheral vascular disease. The date of admission was used as the date of the event. Data on mortality (including cause of death) were retrieved from Statistics Netherlands and coded according to the 10th revision of the International Classification of Diseases.26 Follow-up data on hospitalisation for CV morbidity were derived from records held by PRISMANT, the Dutch national registry of hospital discharge diagnoses.27
By design, the PREVEND study overselected subjects with an elevated UAE to acquire sufficient subjects with microalbuminuria. It should be clear that this is not a random sample of a general population where all elementary units have an equal probability of being selected. Statistical formulae to calculate population parameter estimates should be used to account for the likelihood of selection. A design-based analysis was performed to overcome this overselection of subjects with elevated UAE. This statistical weighting method allows conclusions to be generalised to the general population.28 p Values for trend were calculated between quintiles of MR-proADM. Because of skewed distribution, NT-proBNP, UAE, cystatin C, serum triglycerides and hsCRP were transformed to their natural logarithms. The results are summarised as hazard (risk) ratios (HR) with 95% CIs. To assess which factors are most strongly associated with MR-proADM, we first performed univariate linear regression analyses, followed by multivariate backward linear regression analyses using bootstrapping (1000×, entry criterion 70%; default p value for model entry <0.05, default p value to remain in model <0.10). To avoid multi-colinearity in the latter analyses, we selected the strongest variable from strongly related domains (waist from the waist/BMI domain, systolic blood pressure from the blood pressure/hypertension domain, cystatin C from the renal domain). Kaplan–Meier estimates of the distribution of times from baseline to CV events were generated; log-rank tests were calculated to compare the survival curves between the groups. Multivariate Cox proportional hazards regression analysis was performed, with all significant parameters from the univariate analysis and other relevant covariates from previous studies. MR-proADM was entered linear and log-transformed to assess best fit. We observed a significant statistical interaction between log-transformed MR-proADM and age and CV event outcome. Therefore, an interaction variable was added to the model and HRs derived from this Cox proportional hazards model were plotted. Subjects were classified as young, middle-aged and old (30, 50 and 70 years, respectively) to assess interpretation of clinical value. To assess the additive value of MR-proADM over the FRS, we evaluated the Intergrated Discrimination Improvement and Net Reclassification Improvement (NRI) indices for MR-proADM (as a continuous variable) according to FRS (divided into risk categories). Subjects with a history of CV disease at baseline were excluded from the analysis with the FRS. We evaluated the NRI in the entire population and in age strata because of an interaction with age. All reported probability values are two-tailed and p<0.05 was considered statistically significant. All analyses were performed using StataIC (V.11.0 software for Windows).
The frequency distribution of MR-proADM is shown in figure 2. The distribution appears normal with a few positive outliers and 90% of measurements below 0.55 nmol/l. The mean±SD MR-proADM at baseline was 0.39±0.14 nmol/l, mean age was 49±13 years and 49% were men. The baseline characteristics for all subjects and for subjects classified by quintiles of MR-proADM are summarised in table 1. Compared with lower quintiles, individuals in higher quintiles were significantly older, had higher BMI, cholesterol and glucose levels, higher blood pressure and heart rate and suffered more from CV disease at baseline (all p for trend <0.001). hsCRP, UAE, cystatin-C and NT-proBNP levels were also significantly elevated in higher quintiles (all p for trend <0.001).
In univariate analysis, all investigated subject characteristics except HDL-cholesterol, eGFR and female gender correlated positively with MR-proADM, with strongest associations for age (R2=0.20), eGFR (R2=0.10), cystatin C (R2=0.09), waist circumference (R2=0.09), NT-proBNP (R2=0.08), BMI (R2=0.07), blood pressure (R2=0.07) and hypertension (R2=0.07) (table 2). In a backward multivariate analysis, age, waist circumference, heart rate, NT-proBNP, cystatin C, smoking status and presence of diabetes mellitus remained significantly associated with MR-proADM (model R2=0.28, table 2).
A total of 7903 subjects were followed for a median of 10.5 years (IQR 9.9–10.8). The pre-specified primary endpoint occurred in 752 subjects (9.5%). The incidence of CV events increased with increasing quintiles of MR-proADM, from 8.0% in the bottom quintile (<0.21 nmol/l) to 44.4% in the top quintile (>0.59 nmol/l) (p<0.001 for trend). A Kaplan–Meier analysis of time to first CV event according to quintiles of MR-proADM is shown in figure 3.
In total, 576 subjects (7.3%) died during follow-up, 145 (25.2%) of CV causes. All-cause mortality increased with increasing quintiles of MR-proADM from 7.6% in the bottom quintile to 48.1% in the top quintile (p<0.001 for trend). The incidence of CV-related mortality also increased with higher levels of MR-proADM (p<0.001).
In Cox proportional hazard regression analyses, log-transformed MR-proADM was significantly associated with an increased risk of CV events in both crude models and models adjusted for Framingham CV risk factors (age, gender, blood pressure, HDL cholesterol, diabetes mellitus and smoking) and other CV markers (NT-proBNP, hsCRP and UAE). The latter was not the case for all-cause mortality. A significant interaction between MR-proADM and age was found for incident CV events (p=0.002). The adjusted increase in predicted hazard for CV risk with higher MR-proADM obtained from the Cox proportional hazard analysis is depicted against age in figure 4. A subject of mean age (50 years) and mean MR-proADM (0.39 nmol/l) was used as a referent. The figure shows that, in older subjects (70 years), variation in MR-proADM levels is not associated with CV disease risk, unlike middle-aged (50 years) and younger subjects (30 years). In a middle-aged subject there is an almost linear relationship between MR-proADM and CV risk. In a younger subject, the relationship with increase in CV risk is closer to exponential.
The Integrated Discrimination Improvent index for the model (the same model used in the Cox proportional hazard regression analysis) including MR-proADM was significant, p=0.002 (outcome variable: CV mortality and morbidity). After excluding subjects with a previous CV history, the area under the curve for the FRS in our population is 82% for predicting CV events. Adding MR-proADM to the FRS improved the area under the curve by 0.5% (p<0.001; c-statistics). In the entire population, the NRI of MR-proADM over the FRS was nearly significant (p=0.08) and was mainly driven by correct reclassification to a lower risk category in 4% of subjects who did not have a CV event (p<0.001). Given the interaction between MR-proADM and age for CV risk, we repeated the NRI analysis in subjects <70 years of age (N=7475). This resulted in reclassification of 413 subjects (p=0.003) and we observed a significant upward reclassification in 41 subjects with a CV event (p=0.017), as well as a significant downward reclassification in 209 subjects without a CV event (p<0.001; table 3).
This study describes the association between plasma MR-proADM and CV event risk and outcome in the general population. MR-proADM appears to be a strong independent predictor of CV events, particularly in younger subjects (≤70 years), and adds to the existing conventional and novel CV risk marker prediction models. The large PREVEND cohort, with almost 83 000 subject-years of follow-up, provides a good opportunity for large-scale evaluation of this emerging biomarker and gives new insights into its association with CV disease.
Reliable quantification of ADM has been hampered by its short half-life, the immediate binding of ADM to receptors in the vicinity of its production site and technical difficulties.9 ,29 Limited general population data are available on the stable equivalent of ADM, MR-proADM. Smith et al10 and Melander et al30 measured MR-proADM in a general population cohort and found mean±SD levels of 0.42±0.13 nmol/l and 0.46±0.13 nmol/l, respectively. Bhandari et al examined hypertensive subjects and found much higher levels of MR-proADM, with a mean±SD of 0.59±0.18 nmol/l in subjects without left ventricular hypertrophy and 0.73±0.25 nmol/l in subjects with left ventricular hypertrophy.16 In another study by Dieplinger et al, MR-proADM was assessed in a population with mild to moderate renal dysfunction and found higher levels of MR-proADM, increasing with worsening renal function,7 from 0.43 to 1.34 nmol/l in subjects with eGFR >90 and <30, respectively. All studies—including ours—found a relatively normal distribution of MR-proADM with only a small number of outliers at the high end of the distribution curve. The mean±SD MR-proADM level in our population was 0.39±0.14 nmol/l, lower than in the studies mentioned above. This difference may be explained by the fact that the PREVEND cohort is comprised of relatively young subjects (mean age 49 years) with a low prevalence of comorbidities at baseline. In line with other studies, the strongest correlations were found with age and kidney function in both univariate and multivariate analyses. Waist circumference, BMI and blood pressure were also correlated with MR-proADM in our cohort but, in the multivariate model, only waist circumference remained significantly associated. MR-proADM may be most affected by ageing and kidney function, but may also be influenced by pressure and volume load, as reflected by the independent association with blood pressure and NT-proBNP.
We evaluated the association between MR-proADM and CV events (ie, combined CV mortality and morbidity). The highest event rates were found in the higher range of plasma MR-proADM values (5th quintile: ≥0.59 nmol/l). The incidence of CV events followed a logarithmic increase for higher values of MR-proADM, indicating that subjects with the highest levels of MR-proADM are most at risk for CV disease. This effect remained when adjusted for all relevant CV risk variables.
In crude unadjusted models, the predictive value of MR-proADM for all-cause and CV mortality and CV morbidity is high. Interestingly, when adjusted for common variables such as the conventional CV risk factors (FRS) and several emerging CV risk factors including hsCRP and NT-proBNP, the predictive value of MR-proADM increased for CV events. MR-proADM was not an independent predictor for all-cause mortality after adjusting for the CV risk factors mentioned above. For CV mortality, the predictive value of MR-proADM on outcome did not reach statistical significance (p=0.33), but this may be due to lack of power given the limited number of fatal CV events.
Survival analyses showed a significant interaction between MR-proADM and age for prediction of CV events, which is consistent with the strongest association with the primary endpoint in younger subjects. NRI analyses resulted in a significant reclassification in middle-aged and young subjects, with subjects with a CV event correctly reclassified into a higher risk category and event-free subjects into a lower risk category. This reclassification was not present in subjects >70 years (p=0.32), suggesting added prognostic value for younger subjects in particular. The reason for this interaction with age is unknown. A possible explanation may be that, in older subjects, upregulation of ADM is at the end of its dose-effect relationship while correlation with outcome is visible primarily in early subclinical stages of CV disease.
CV risk stratification using biomarkers can help to identify subjects in the community who may benefit most from preventive therapeutic interventions. In populations with a low to intermediate risk for CV disease, data on the additive value of established and/or novel biomarkers are conflicting. Our results are in agreement with previous publications regarding the prognostic performance of MR-proADM. Most data were obtained in different cohorts with comorbidities,11–13 ,15 but also in the community.30 Our results provide valuable knowledge about MR-proADM as an effective predictive biomarker for future CV events in younger subjects without other comorbidities or a history of CV disease. Whether specific preventive strategies or treatment may be of benefit for subjects with increased MR-proADM remains to be addressed before this biomarker can be used for routine screening.
This study gives a detailed overview of the distribution of MR-proADM in the general population and provides evidence for the value of MR-proADM as a potent and interesting biomarker for predicting CV events. We postulate that MR-proADM may be particularly valuable as a biomarker in younger subjects.
Competing interests JS is employed by BRAHMS GmbH/ThermoFisher Scientific, a company which manufactures and holds patent rights on the MR-proADM assay.
Ethics approval This study was conducted with the approval of the local ethics committees and performed in accordance with the Declaration of Helsinki.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.