The INVOR-Study [19] (Study of Incident Dialysis Patients in Vorarlberg) is a single-center, prospective, observational cohort study of incident Caucasian hemodialysis and peritoneal dialysis patients in Vorarlberg, the westernmost province of Austria counting approximately 400,000 inhabitants.

Ethic statement: The study was approved by the ethics committee of the Innsbruck Medical University and all patients enrolled in the study provided written informed consent.

All incident dialysis patients from this province starting chronic dialysis treatment between May 1^st, 2000 and April 30^th, 2006 were consecutively enrolled with the advantage that all patients of this region are treated by the same care provider. During this period of 6 years a total number of 235 incident dialysis patients were included and followed until the study endpoint was reached or follow-up was censored at December 31^st, 2009. Ten patients having a malignant tumor at initiation of dialysis were not recruited defined by the exclusion criteria. Due to inappropriate or missing blood samples MR-proANP and MR-proADM was measured in 201 out of 235 patients. All data and analyses described in this manuscript are based on these 201 patients.

Clinical, laboratory and medication data were collected prospectively starting at the time of initiation of dialysis. These data included age, sex, height, weight, body mass index, diabetes status and current smoking status. Type of and change in renal replacement therapy (hemodialysis, peritoneal dialysis and kidney transplantation) were recorded and considered as time-dependent treatment status for data analysis. Vascular access procedures and the type of vascular access (native fistula, graft or central venous catheter) for hemodialysis were also evaluated.

Information on the following clinical events were collected before initiation of dialysis and during the entire observation period thereafter: coronary artery disease (including myocardial infarction, percutaneous transluminal coronary angioplasty, aortocoronary bypass), cardiovascular disease (including myocardial infarction, percutaneous transluminal coronary angioplasty, aortocoronary bypass, angiographically-proven coronary stenosis ≥50%, sudden cardiac death, ischemic or hemorrhagic cerebral infarction, transient ischemic attack, carotid stenosis and carotid endarterectomy), peripheral arterial disease (significant ultrasound- or angiographically-proven vascular stenosis, percutaneous transluminal angioplasty, peripheral bypass, amputation).

We used two novel commercially available fully automated sandwich immunoassays for the measurement of MR-proANP (B.R.A.H.M.S MR-proANP KRYPTOR) and MR-proADM (B.R.A.H.M.S MR-proADM KRYPTOR) according to the manufacturer's instruction manuals (B.R.A.H.M.S GmbH, Hennigsdorf Germany). The design of these assays is based on immunoluminometric assays described previously [20], [21]. Both parameters were measured in plasma collected at the time immediately before the initiation of the first dialysis therapy and kept frozen at −80°C until measurement in a single batch. Other laboratory parameters reported here were measured from the same blood collection immediately before the first dialysis therapy.

As described earlier, the limit of quantitation was 4.5 pmol/L for the MR-proANP assay with a within-run imprecision coefficient of variation (CV) of <4.5% between 10 and 20 pmol/L and <2.5% between 20 and 1000 pmol/L, and between-run imprecision CV of <6.5% between 10 and 1000 pmol/L [21]. The limit of quantification was 0.23 nmol/L for the MR-proADM assay with a within-run imprecision CV of <4% between 0.5 and 2 nmol/L and <2% between 2 and 6 nmol/L, and a between-run imprecision CV of <11% between 0.5 and 2 nmol/L and <10% between 2 and 6 nmol/L [20]. All blood samples were processed by personnel blinded from any patient data.

The first outcome of interest was all-cause mortality and cardiovascular mortality. We also investigated other endpoints such as cardiovascular disease and peripheral arterial disease. Cardiovascular mortality was defined as death of myocardial infarction, heart failure, sudden death, ischemic or hemorrhagic stroke. Cardiovascular disease (CVD) events were defined as fatal and non-fatal myocardial infarction, percutaneous transluminal coronary angioplasty (PTCA), aortocoronary bypass (ACBP), angiographically-proven coronary stenosis ≥50%, ischemic cerebral infarction or transient ischemic attack. For peripheral arterial disease (PAD) at least one of the following events was existent: significant ultrasound- or angiographically-proven vascular stenosis, percutaneous transluminal angioplasty (PTA), peripheral bypass or amputation. An incident PAD event was only considered as a first time manifestation or a deterioration of PAD in terms of e.g. a change in PAD stage according to Fontaine. Two patients were lost to follow-up, one regained renal function and the other one moved away.

At baseline, categorical data were compared using χ²-test and continuous variables were analyzed using an unpaired t-test or the non-parametric Wilcoxon rank-sum test. Data are presented as mean±SD and as median and 25^th and 75^th percentiles for skewed variables where appropriate.

To investigate the influence of MR-proANP and MR-proADM on all-cause mortality, cardiovascular mortality, cardiovascular disease and peripheral arterial disease, multivariable adjusted Cox-proportional hazards regression models were performed. In all analyses a p-value of 0.05 was considered significant. Variables were chosen for the multiple Cox regression analysis since they were either well-known risk factors for all-cause or CV mortality or they showed correlations with MR-proADM or MR-proANP. The first model was adjusted for age, sex, previous cardiovascular events, diabetes mellitus and type of renal replacement therapy, which was modeled time-dependently (Model 1). The second model (Model 2) was additionally adjusted for albumin, C-reactive protein, current smoking status, native fistula and echocardiographic data (ejection fraction ≤60% versus >60%). Due to sparseness of the data and risk of overfitting, this fully-adjusted model should only be considered as a sensitivity analysis to Model 1, which is taken as the main model in this investigation. Linear relationship assumption of MR-proANP and MR-proADM in the Cox models was tested, as well as the proportional hazards assumption. One standard deviation (SD) was taken as the unit of measure for each of the continuous outcome variables to ensure comparability of Hazard Ratios. For ease of interpretation, additional fully adjusted Cox-proportional hazards regression models were calculated by dividing MR-proANP and MR-proADM into tertiles and also by stratifying patients in groups of high risk (both MR-proANP and MR-proADM in the upper tertile), intermediate risk (only one of the two parameters in the upper tertile) and low risk (none of the two parameters in the upper tertile). Adjusted survival curves are given for each of these risk groups and for each of the tertiles, holding all covariates fixed at their mean level. We also performed a sensitivity analysis with censoring at the time of transplantation. All analyses were conducted in SPSS version 16.0 software (SPSS Inc., Chicago, IL, USA) and R using the package “survival”.

Table 3 presents the results from Cox-proportional hazards models for MR-proANP and MR-proADM, and all-cause and cardiovascular mortality as well as CVD and PAD events. Hazards ratios are adjusted for age, sex, previous CVD events, diabetes mellitus and time-dependent type of renal replacement therapy (Model 1) and an extended adjustment additionally for albumin, CRP, current smoking, native fistula and ejection fraction (Model 2 as sensitivity analysis). On a continuous scale, both MR-proANP and MR-proADM were significantly associated with all-cause and with cardiovascular mortality but not with the entire group of fatal and non-fatal cardiovascular disease events. When PAD was used as outcome variable only plasma MR-proANP but not MR-proADM levels showed a significant association. These results were only marginally influenced if further adjusted for albumin, CRP, current smoking, native fistula and ejection fraction determined by echocardiography (Model 2).

When we analyzed the data according to tertiles of MR-proANP and MR-proADM (Table 3), a non-significant trend for the upper tertile of MR-proANP for all-cause (HR = 1.76, p = 0.082) and cardiovascular mortality (HR = 2.96, p = 0.053) was observed. However, patients in the upper tertile of MR-proADM had a significantly increased risk for all-cause (HR = 2.39, p = 0.003) as well as cardiovascular mortality (HR = 3.16, p = 0.013). Survival curves for all-cause and cardiovascular mortality demonstrated that the first and second tertile of MR-proANP and MR-proADM were similar and only the third tertile discriminated between survivors and non-survivors (Figure 2).

This increased risk was clearly identified using a stratification based on the tertiles of the two peptides by combining the highest tertiles of MR-proANP (≥794 pmol/L) and MR-proADM (≥3.11 nmol/L). Table 4 provides the demographic and clinical characteristics of patients stratified in high risk (both MR-proANP and MR-proADM in the upper tertile), intermediate risk (only one of the two parameters in the upper tertile) and low risk (none of the two parameters in the upper tertile). The increase in risk with respect to all-cause mortality was almost identical when we compared the high risk group versus intermediate risk group (HR = 2.70, p = 0.001) and the high risk group versus low risk group (HR = 3.02, p<0.001). The risk was only increased when a patient had both parameters in the upper tertile, although demographic, clinical and laboratory variables were similar among the intermediate and high risk group. The high risk group was associated with an about 3-fold higher risk for all-cause mortality (HR = 2.87, p<0.001) and cardiovascular mortality (HR = 3.58, p = 0.001) when compared to the remaining patients (Table 3 and Figure 2).

It might introduce some considerable bias to the data analysis when a patient is selected for transplantation and follow-up time is censored at the time of transplantation. For this reason the above main analysis was performed by a time-dependent modeling of the renal replacement therapy status as discussed earlier [22], [23]. To exclude that this procedure has influenced our main findings, we performed a sensitivity analysis with classical censoring at the time of transplantation which did, however, not reveal any substantial differences in HRs compared to the primary analysis (see Table S1). Due to sparseness of the data and risk of overfitting we also calculated a simple model only adjusting for age and sex that did not show major differences in estimates compared to Model 1. A further sensitivity analysis was calculated where Model 1 was additionally adjusted for hemoglobin. No substantial differences in estimates could be observed.

The prospective recruitment of all patients starting dialysis treatment over a period of six years in a clearly described area allowed a complete recruitment of patients requiring renal replacement therapy with almost no loss to follow-up during a long observation period. We can therefore exclude the most important bias of cross-sectional studies with a mix of prevalent and incident cases and the resulting survival bias. We furthermore considered even the observation period following kidney transplantation by modeling the treatment status in a time-dependent model, but on the other hand excluded by sensitivity analysis, that this procedure has obscured our results.

There are also some limitations of this study. A first limitation is the relatively small number of patients and events. We therefore could only adjust our analyses for a small number of variables (6 variables in Model 1). An extended adjustment with more variables (Model 2 and additional sensitivity analyses) was provided but should only be considered as sensitivity analysis. Interestingly, the estimates for MR-proADM and MR-proANP remained widely stable in Models 1 and 2, the additional sensitivity analyses and a very simple model only adjusted for age and sex. This argues for a pronounced independence of the two investigated variables for the prediction of endpoints. Second, we examined patients at the time immediately before renal replacement therapy was started. Therefore, some patients might not have been in a steady metabolic or extracellular volume state of the parameters evaluated. Concentrations of these hormones might therefore not necessarily be applicable to other cohorts of patients with CKD. On the other hand, values measured after the start of dialysis treatment as done in other mostly smaller studies might represent dialysis dose or dialysis efficacy and thus answer a different question. Third, our study includes only Caucasian patients of a clearly described geographical region with almost complete ascertainment of incident dialysis patients over a defined period of time. It therefore lacks generalizability to other ethnic populations as well as other recruitment procedures. Finally, since the measurements of the two peptides are not rigorously standardized, concentrations measured by various assays are not necessarily comparable.

Among incident dialysis patients MR-proANP and MR-proADM were shown to be associated with all-cause mortality and cardiovascular mortality, with the highest risk when both parameters were in the upper tertiles.