Pro-Adrenomedullin predicts 10-year all-cause mortality in community-dwelling patients: a prospective cohort study

Background Several studies found mid-regional pro-adrenomedullin (ProADM), the prohormone of the cardiovascular protein adrenomedullin, to be strongly associated with short-term mortality, mostly in the inpatient setting. We evaluated associations of ProADM levels with 10-year mortality in community-dwelling primary care patients with respiratory tract infections. Methods This is a post-hoc analysis using clinical and biomarker data of 134 primary care patients with respiratory tract infections. ProADM was measured on admission and after 7 days in batch-analysis. 10-year follow-up data was collected by GP, patient and relative tracing through phone interviews. We calculated Cox regression models and area under the receiver operating characteristics curves to assess associations of ProADM with 10-year all-cause mortality. Results During the 10-year follow-up 6% of included patients died. Median baseline ProADM blood levels (nmol/l) were significantly higher in non-survivors compared to survivors (0.5, IQR 0.4–1.3; vs. 0.2, IQR 0.1–0.5; p = 0.02) and showed a significant association with 10-year all-cause mortality in an age-adjusted cox regression model (HR: 2.5, 95%-CI: 1.0–6.1, p = 0.04). ProADM levels on day 7 showed similar results. Conclusions This posthoc analysis found an association of elevated ProADM blood levels and 10-year all-cause mortality in a primary care cohort with respiratory tract infections. Due to the methodological limitations including incomplete data regarding follow-up information and biomarker measurement, this study warrants validation in future larger studies. Trial registration Current Controlled Trials, SRCTN73182671


Background
Prognostic blood biomarkers have generated clinical interest, resulting in therapeutic and diagnostic improvements [1]. A more personalized and targeted medicine offers opportunities to enhance treatment safety, efficacy as well as cost effectiveness. In this context, mid-regional proadrenomedullin (ProADM) provides prognostic information in various clinical settings and across different patient populations, predominantly suffering from cardiovascular and infectious diseases [2][3][4][5][6][7][8].
Stratification of patients with lower respiratory infection based on a clinical algorithm including ProADM stratification tended to shorten length of hospital stay without an increase in adverse clinical outcome [41,42].
In the setting of primary care, in smaller studies ProADM was associated with mortality in patients with urinary tract infections [43], type 2 diabetes [43][44][45] or heart failure [25,[46][47][48][49]. It is not known whether this marker helps to predict long-term risk, which could be useful for directing preventive measures. The aim of the current analysis was to investigate the ability of ProADM to predict long-term all-cause mortality and adverse outcome in a primary care cohort, visiting their general practitioner for respiratory infections.

Study design and setting
This is a post-hoc analysis investigating ProADM in a primary care cohort enrolled between December 2004 and April 2006 into the PARTI intervention trial (Procalcitonin-Guided Antibiotic Use vs. a Standard Approach for Acute Respiratory Tract Infections in Primary Care) [50,51]. In brief, this randomized, multicenter, noninferiority trial investigated the feasibility of a PCT-guided antibiotic therapy visiting their GP for a respiratory infection. The use of antibiotics was more or less strongly discouraged based on defined PCT-cutoff ranges. The standard approach followed evidence-based guidelines for use of antibiotics. The aim of the trial was to investigate safety and efficacy of using PCT levels to guide antibiotic therapy.
The initial study protocol, as well as the present follow-up trial, were approved by the local Ethics Committee of Basel (EKBB). Written informed consent was obtained from all participating physicians and patients.

Selection of participants
Initially, all patients with the diagnosis of upper or lower ARTI and the physician's intention to prescribe antibiotics were included. Exclusion criteria were antibiotic use within the previous 28 days, psychiatric disorders or inability to give written informed consent, not being available for follow-up, not being fluent in German, severe immunosuppression, cystic fibrosis, active tuberculosis, and the need for immediate hospitalization. In this analysis, we included only patients, of which the 10-year follow-up and either baseline or day 7 ProADM levels were available.

Data collection and endpoints
The primary endpoint was defined as long-term 10-year all-cause mortality. Secondary outcomes were adverse outcomes including death, pulmonary embolism, and major adverse cardiac or cerebrovascular events (MACCE), which includes cardiac infarction, cardiac arrest, stroke and transient ischemic attack. In addition, we investigated new onset of diabetes within the follow-up period.
To verify outcomes, we performed follow-up telephone interviews with patients, relatives and/or general practitioners (GPs) 10 years after the baseline visit, using systematic questionnaires. Also, the register of death of the cantons Basel-Stadt and Basel-Land was consulted if no information about vital status was available.

Analysis of blood biomarkers
Blood samples were collected in the primary care centre from each patient on admission and after 7 days in ethylendiaminetetraacetic (EDTA) tubes and sent by courier to the central Laboratory of the University Hospital Basel for measurement of PCT. Leftover blood samples were frozen and stored at −80°for the later measurement of prognostic markers.
ProADM serum values were determined using a sandwich immunoassay with a functional interassay precision of 20% and an assay sensitivity assessed as being 0.  [20].

Statistical analysis
We used STATA 12.1 (STATA Corp, College Station, TX; USA) and created univariable and bivariable (adjusting for age) as well as a multivariable model (adjusting for age, randomisation arm (PCT group), smoking history) cox regression analysis to calculate hazard ratios (HR) and area under the receiver operating characteristics curve (AUC) to investigate the predictive accuracy of Pro-ADM. We used a natural logarithm (base e) transformation of all biomarker data before entering into the statistical models to approximate normal distribution. Therefore, HR corresponds to a 2.72-fold increase in log transformed biomarker levels. To illustrate the predictive power we used Kaplan-Meier plots for proportion of survivors and proportion of events by quartiles of biomarker levels. Log-rank tests were performed to compare quartiles. P-values of <0.05 were considered to indicate statistical significance.

Patient population
The initial cohort included 458 adult patients with an ARTI, of which 167 (36.5%) had ProADM blood samples available, while the other patients had not enough leftover sample for measurement of ProADM. A total of 134 (80.2%) of these patients could be contacted to assess long-term outcomes between April and August 2015 and were, thus, included in the final analysis. For 291 (63.5%) patients of the initial cohort no data concerning ProADM blood levels was available due to missing blood tube. Table 1 shows baseline characteristics of the overall cohort as well as stratified by survival status. The median age was 42.0 years and 32.8% of patients were male. Median ProADM blood levels on admission were 0.3 nmol/l and 0.2 nmol/l on day 7. According to survival status, there were significant differences in age, comorbidities (chronic obstructive pulmonary disease), initial clinical classification of type of respiratory infection and nicotine consumption (pack-years). A comparison between the initial cohort and the patients available for 10-year follow-up is presented in Table 5 (see Appendix).
A similar 10-year post-hoc analysis of the prognostic biomarkers copeptin and MR-proANP, based on the same blood samples from the PARTI trial, has been published recently and showed that copeptin as well as MR-proANP were associated with 10-year all-cause mortality [52,53].
Primary outcome: 10-year all-cause mortality During the follow-up of 10 years (mean: 9.5 years), mortality was 6% (n = 8 patients). Median admission ProADM blood levels (nmol/l) were significantly higher in non-survivors compared to survivors (0.5, IQR 0. Initial ProADM level showed a strong association with 10-year all-cause mortality in a univariable cox regression model (HR: 4.2, 95%-CI: 1.5, 11.5, p = 0.006). The result remained significant when adjusted for age (adjusted HR: 2.5, 95%-CI: 1.0-6.1, p = 0.043). There was no significant effect modification by randomisation arm (PCT group) or a positive history of smoking. Day seven ProADM levels showed in the univariable model a  Arterial hypertension (%) 13 (9.7) 12 (9.5) 1 (12.5) 0.78 We generated Kaplan-Meier curves to visualize the difference of survival between the highest quartile and lower three quartiles on admission and on day 7 (Figs. 1  and 2). The log-rank-test showed no significant increase in mortality when comparing the highest ProADM quartile to quartiles 1 to 3.
Cox regression models found borderline significant associations between initial and ProADM levels and adverse outcome, with an AUC of 0.62 (Table 2). When adjusted for age (HR: 1.4, 95%-CI: 0.8 to 2.5), p = 0.273), there was no significant association found. As well, there was no significant effect modification by randomisation arm (PCT group) or a positive history of smoking.
Significance tests for follow-up ProADM levels showed similar results.
Kaplan-Meier curves (Figs. 3 and 4), with patients stratified based on ProADM quartiles, illustrate the slightly higher event rates in patients in the highest ProADM quartile.
Further, no association between ProADM and MACCE or new-onset diabetes mellitus was found (see Appendix).

Discussion
Within this prospective, observational 10-year follow-up study of a small cohort of community-dwelling ARTI patients, we found ProADM on admission and after 7 days to be an age-independent predictor for all-cause mortality. No significant associations of ProADM and secondary outcomes were found, namely adverse outcome, MACCE and new onset of diabetes.
Although there is no clear understanding why an increase in ProADM points to increased mortality risk, existing data suggests that elevated ProADM levels reflect disease severity and endothelial and cardiovascular dysfunction [11][12][13][14][15][16][17][18][19]. Also, higher ADM levels increases cardiac output, induces hypotension and vasodilation, and increases glomerular filtration rate and fractional sodium excretion [10,19,54], thereby inducing a reduction in cardiac pre-and afterload [20]. Thus, the involvement of ADM in several pathological disease states and  Median ProADM blood levels in our outpatient cohort were 0.3 nmol/l on admission and 0.2 nmol/l on day 7 and thus significantly lower compared to other hospitalized patient cohorts. The AtheroGene study found median ProADM levels of around 0.5 and 0.6 nmol/L in patients with stable angina and acute coronary syndrome [35]. The LAMP study reported median concentration of 0.73 nmol/L in patients with myocardial infarction [36], while the GISSI study found a median ProADM concentration of 0.75 nmol/L in patients with chronic heart failure [26]. Analysis of a presumably healthy subset in a large outpatient cohort (n = 5258) lead to a reference interval of 0.23-0.64 nmol/L [18,55]. These differences demonstrate that levels of ProADM need to be adapted to the specific clinical setting to be interpreted in a meaningful way.
Interestingly, although several previous studies have suggested that ProADM was also associated with other adverse outcomes in addition to all-cause mortality [26,27,[33][34][35][36][44][45][46], we did not find such statistically significant association with the incidence of our secondary combined endpoint including pulmonary embolism and MACCE. Also, in contrast to another study [39], our analysis did not find an association of ProADM with new onset diabetes mellitus. This may be due to the small number of events in our generally healthy population with a low burden of comorbidities and thus low power of our analysis. Further, the respiratory infection of patients during the initial trial may have had an influence on ProADM levels. Thus, similar to lipid levels,   [56] this marker may be best analyzed during stable conditions for the purpose of long-term risk assessment.
The main strengths of this study include the 10 years of follow-up, the participation of multiple GP practices, and the community sample of patients with ARTI of different severity representative for patients mainly treated in primary care. Nonetheless, we are aware of several limitations. First, this is a secondary analysis of a previous trial and baseline risk assessment is incomplete as is the availability of ProADM levels in the cohort. For 291 (63.5%) patients no data concerning ProADM blood levels was available, because blood sampling was only done in a sub fraction of the overall cohort during a certain time period. Selection bias is thus possible. Second, due to the long follow-up period, a recall bias has to be considered. Further, no information was available on the cause of death, when patients were tracked through the register of deaths. Third, our sample was small and we observed only few events for the analysis of the relationship between ProADM levels and adverse outcomes.

Conclusion
This posthoc analysis found an association of elevated ProADM blood levels and 10-year all-cause mortality in a primary care cohort with respiratory tract infections. Due to the methodological limitations including incomplete data regarding follow-up information and biomarker measurement, this study warrants validation in future larger studies. If validated, ProADM may help to risk stratify patients and thereby allows to improve allocation of health care resources and preventive measures. Appendix T3 T4 T5

Availability of data and materials
The data set supporting the results of this article is included within the article and its additional files.

Authors' contributions
Mr. JO and Mr. MM contributed equally to this study. Mr. JO, Mr. MM and Dr. PS had full access to all of the data in the study and take responsibility for the integrity of the data. Mr. JO, Mr. MM and Dr. PS performed the statistical work and drafted the manuscript. All authors helped to interpret the findings, read and revised the manuscript critically for important intellectual content, and approved the final version of the manuscript.

Ethics approval and consent to participate
The initial study protocol, as well as the present follow-up trial, were approved by the local Ethics Committee of Basel (EKBB). Written informed consent was obtained from all participating physicians and patients.

Consent for publication
Not applicable. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Drs. Christ-Crain, Kutz, Mueller and Schuetz received support from BRAHMS to attend meetings and fulfilled speaking engagements. Drs. Schuetz, Christ-Crain and Mueller received support from bioMérieux to attend meetings and fulfilled speaking engagements. Dr. Mueller has served as a consultant and received research support from BRAHMS and bioMérieux. All other authors declare that the answer to the questions on the competing interest form are all "No" and therefore have nothing to declare. All authors confirm that they do not have a conflict of interest associated with this manuscript.