Copeptin Levels Associate with Cardiovascular Events in Patients with ESRD and Type 2 Diabetes Mellitus

Abstract

In ESRD, the neurohormone arginine vasopressin (AVP) may act primarily through V1a and V1b receptors, which promote vasoconstriction, myocardial hypertrophy, and release of adrenocorticotropic hormone. The preanalytical instability of AVP limits the investigation of whether this hormone associates with cardiovascular events, but the stable glycopeptide copeptin may serve as a surrogate because it is co-secreted with AVP from the posterior pituitary. Here, we studied whether copeptin predicts cardiovascular risk and mortality in ESRD. We measured copeptin at baseline in 1241 hemodialysis patients with type 2 diabetes participating in the German Diabetes and Dialysis Study. The median copeptin level was 81 pmol/L (interquartile range, 81 to 122 pmol/L). In Cox regression analyses, compared with patients with copeptin levels in the lowest quartile (≤51 pmol/L), patients with copeptin levels in the highest quartile (>122 pmol/L) had a 3.5-fold increased risk for stroke (HR, 3.48; 95% CI: 1.71 to 7.09), a 73% higher risk for sudden death (HR, 1.73; 95% CI: 1.01 to 2.95), a 42% higher risk for combined cardiovascular events (HR, 1.42; 95% CI: 1.06 to 1.90), and a 48% higher risk for all-cause mortality (HR, 1.48; 95% CI: 1.15 to 1.90). In contrast, we did not detect significant associations between copeptin levels and risks for myocardial infarction or death caused by congestive heart failure. In conclusion, copeptin levels strongly associate with stroke, sudden death, combined cardiovascular events, and mortality in hemodialysis patients with type 2 diabetes. Whether vasopressin receptor antagonists will improve these outcomes requires further studies.

Arginine vasopressin (AVP) is a key neurohormone in the human body, with numerous important physiologic activities including regulation of BP, fluid volume, and serum osmolality. AVP acts via three G protein–coupled receptors, namely the V1a (vasoconstriction and myocardial hypertrophy), the V1b (release of adrenocorticotropic hormone), and the V2 receptor (water retention in the renal collecting duct). Despite its suggested importance in the pathogenesis of cardiovascular disorders,^1,2 the evaluation of AVP secretion is difficult because of its preanalytical instability. Copeptin, a 39-amino acid glycopeptide that comprises the C-terminal part of the AVP precursor, is co-secreted with AVP from the neurohypophysis and is a stable and sensitive surrogate marker for AVP release, making it much easier to measure.³

Dialysis patients are at a very high risk of death (about 20% per year)⁴. Thus, the identification of new risk factors with the potential of applying intervention strategies to improve outcome in the future is of utmost importance.

It is assumed that AVP in ESRD patients with little or no residual urine output can probably not efficiently act via V2 receptors. Evidence suggests that downregulation of V2 receptor mRNA and a deficient AVP-stimulated adenyl cyclase result in resistance of V2 receptors in earlier stages of chronic kidney disease.^5–7 These findings suggest that, in ESRD, AVP might primarily act via the V1a and V1b receptors. Thus, an increase in cardiovascular risk and all-cause mortality in ESRD patients might be partly linked to the predominant activation of V1a and V1b receptor function.

This study evaluates the association of plasma copeptin concentrations with cardiovascular events and all-cause mortality in hemodialysis patients with type 2 diabetes mellitus (T2DM). To this end, data from the German Diabetes Dialysis Study,⁸ which evaluated atorvastatin in 1255 hemodialysis patients, are analyzed.

RESULTS

Of 1255 patients who took part in the 4D study, 1241 (631 on placebo and 610 on atorvastatin) had a baseline copeptin measurement. The mean follow-up-period was 3.96 years (median, 4.0 years) for patients on atorvastatin and 3.91 years (median, 4.08 years) for those on placebo. During follow-up, 609 of 1241 patients died, of whom 158 patients died of sudden death and 40 because of congestive heart failure (CHF). A total of 462 patients reached the cardiovascular endpoint (CVE), with myocardial infarction (MI) and stroke occurring in 196 and 102 patients, respectively.

Median baseline copeptin was 80.8 pmol/L (interquartile range, 51.2 to 122 pmol/L). Patient baseline characteristics are shown in Table 1. Patients with higher copeptin concentrations were on maintenance hemodialysis treatment for a longer period of time. They showed higher systolic BP values and were treated with higher ultrafiltration volumes. With ascending copeptin, the N-terminal pro B-type natriuretic peptide (NT-pro-BNP) levels were increasing also, and treatment with diuretics became less frequent. Significant correlation coefficients were found for copeptin and time since start of maintenance hemodialysis treatment (correlation coefficient, 0.304; Figure 1A), NT-pro-BNP (0.265) (Figure 1B), ultrafiltration volume (0.232), use of diuretics at baseline (−0.194), body mass index (−0.190), phosphate (mg/L) (0.173), albumin (g/dl) (0.106), systolic BP (0.084), and peripheral vascular disease (0.061).

Table 1.

Patient characteristics according to quartiles of baseline copeptin

	Copeptin				Pa
	Quartile 1 (≤51.2) (pmol/L) (n = 315)	Quartile 2 (>51.2 to ≤80.8) (pmol/L) (n = 311)	Quartile 3 (>80.8 to ≤122) (pmol/L) (n = 311)	Quartile 4 (>122) (pmol/L)(n = 304)
Age (years)	66.0 ± 8.5	65.2 ± 8.0	65.7 ± 8.4	66.1 ± 8.1	0.732
Gender (male/female)	158/157	176/135	169/142	163/141	0.567
Ever smoking (%; n)	37 (116)	45 (140)	39 (122)	40 (121)	0.430
Body mass index (kg/m2)	28.4 ± 5.1	28.3 ± 4.9	27.1 ± 4.7	26.2 ± 4.1	<0.001
Systolic BP (mmHg)	144 ± 23	144 ± 20	146 ± 23	149 ± 21	0.027
Diastolic BP (mmHg)	76 ± 11	75 ± 11	76 ± 11	77 ± 12	0.166
Ultrafiltration volumeb (kg)	2.0 (1.0–3.0)	2.0 (1.0–3.0)	2.0 (2.0–3.0)	2.9 (2.0–3.3)	<0.001
Arterio-venous fistula (%; n)	92 (289)	94 (293)	94 (291)	93 (283)	0.785
Time receiving dialysis (months)	6.2 ± 6.0	7.0 ± 5.8	8.9 ± 7.3	11.1 ± 7.3	<0.001
History ofc
arrhythmia (%; n)	18 (58)	16 (50)	21 (66)	19 (59)	0.486
myocardial infarction, CABG, PCI, or CHD (%; n)	28 (89)	29 (90)	28 (86)	33 (101)	0.441
CHFd (%; n)	33 (105)	34 (105)	37 (116)	38 (115)	0.548
stroke or TIA (%; n)	17 (52)	15 (47)	18 (56)	22 (67)	0.137
peripheral vascular disease (%; n)	37 (116)	49 (153)	48 (150)	45 (136)	0.013
hemoglobin (g/dl)	11.0 ± 1.3	10.9 ± 1.4	10.9 ± 1.4	10.9 ± 1.3	0.630
HbA1c (%)	6.67 ± 1.24	6.73 ± 1.27	6.73 ± 1.26	6.75 ± 1.25	0.808
Phosphate (mg/L)	5.59 ± 1.49	6.07 ± 1.50	6.10 ± 1.61	6.38 ± 1.76	<0.001
Albumin (g/dl)	3.78 ± 0.30	3.82 ± 0.29	3.79 ± 0.32	3.87 ± 0.29	<0.001
LDL cholesterol (mg/dl)	125 ± 32	126 ± 30	124 ± 29	127 ± 29	0.656
CRP (mg/L)	4.7	5.4	4.9	5.1	0.723
median (25th and 75th percentile)	(2.1 to 12.7)	(2.9 to 13.4)	(2.4 to 12.0)	(2 to 11.8)
Nt-pro-BNP (pg/ml)	2233	2865	3657	5176	<0.001
median (25th and 75th percentile)	(909 to 5670)	(1191 to 7468)	(1686 to 10,469)	(2251 to 14,702)
Diuretic use at baseline (%; n)	86 (272)	88 (273)	77 (238)	69 (209)	<0.001

Data are given as mean ± SD or median (interquartile range) for skewed variables. To convert hemoglobin to mmol/L, multiply by 0.62. To convert phosphate to mmol/L, multiply by 0.32. To convert LDL cholesterol to mmol/L, multiply by 0.03. CABG, coronary artery bypass grafting surgery; PCI, percutaneous coronary intervention; CRP, C-reactive protein; TIA, transient ischemic attack; HbA1c, glycated hemoglobin.

^aP value for comparison between groups of patients were derived from a general linear model for continuous variables or logistic regression for categorical variables both adjusted for age and gender, as appropriate.

^bThe ultrafiltration volume was calculated based on body weight before and after dialysis at randomization.

^cTypes of disease and intervention are not mutually exclusive.

^dPredominantly New York Heart Association II.

Figure 1.

Correlation between dialysis vintage, Nt-pro-BNP and copeptin depicted as scatterplot for log-transformed copeptin versus time since start of maintenance hemodialysis treatment (A) and log-transformed Nt-pro-BNP (B). The lines represent the cubic regression equation (time since start of hemodialysis = 17.8903 − 7.884751 × log copeptin + 1.277041 × (log copeptin)² and log NT-pro-BNP = 6.936637 − 0.003649 × log copeptin + 0.065427 × (log copeptin)².

Copeptin and Risk of Outcome

Copeptin at baseline was strongly associated with the risk of stroke (Figure 2A; Kaplan-Meier estimates). By Cox regression analyses (model 1), the relative risk of having a stroke was more than three times higher in patients with copeptin concentrations >122 pmol/L (fourth quartile) compared with those with levels ≤51.2 pmol/L (first quartile; hazard ratio [HR], 3.48; 95% confidence interval [CI], 1.71 to 7.09). This association was slightly alleviated after further adjustment for markers of volume status and residual urine output (model 2; HR, 3.01; 95% CI, 1.44 to 6.33). Per unit increase in log copeptin the risk of stroke rose by 73% (HR, 1.73; 95% CI, 1.25 to 2.39). This association was virtually unchanged after controlling for potential confounders (HR_model1, 1.74; 95% CI, 1.23 to 2.47). Further adjustment for markers of volume status and residual diuresis did slightly reduce this risk (HR_model2, 1.62; 95% CI, 1.11 to 2.36). When analyzing ischemic stroke and fatal stroke separately, similar results were found (data not shown). Because of the limited number of hemorrhagic strokes (n = 13), no separate analysis was performed.

Figure 2.

Copeptin is associated with stroke and sudden death. Kaplan-Meier estimates for time to stroke (A) and sudden death (B) in subgroups of patients according to quartiles of copeptin.

Copeptin at baseline was also associated with the risk of sudden death (Figure 2B; Kaplan-Meier estimates). The risk of sudden death increased by 73% in patients with copeptin concentrations >122 pmol/L (fourth quartile) compared with those ≤51 pmol/L (first quartile; HR_model1, 1.73; 95% CI, 1.01 to 2.95). However, this association was no longer significant after adjustment for markers of volume status and residual urine output (HR_model2, 1.47; 95% CI, 0.84 to 2.57). Per unit increase in log copeptin, the unadjusted risk of sudden death increased by 49% (HR, 1.49; 95% CI, 1.16 to 1.92). This association persisted, at least by trend, after adjustment for potential confounders and additional analysis with markers of volume status and residual urine output (HR_model1, 1.44; 95% CI, 1.10 to 1.91; HR_model2, 1.33; 95% CI, 0.99 to 1.80).

No statistically significant association of copeptin with death caused by CHF was observed. However, the HR for death caused by CHF was 9% higher in patients within the fourth quartile of baseline copeptin compared with patients with copeptin concentrations within the first quartile (HR_model1, 1.09; 95% CI, 0.43 to 2.77; HR_model2, 1.09; 95% CI, 0.41 to 2.89). Per unit increase in log copeptin, the HR increased by about one third (HR_model1, 1.33; 95% CI, 0.75 to 2.36; HR_model2, 1.37; 95% CI, 0.74 to 2.55). Also, no association of copeptin with MI was found. These results did not change when silent MIs were excluded or nonfatal and fatal MIs were analyzed separately (data not shown).

However, the risk of having a CVE increased by 42% in patients with copeptin concentrations >122 pmol/L compared with those with levels ≤51 pmol/L (HR_model1, 1.42; 95% CI, 1.06 to 1.90; HR_model2, 1.29; 95% CI, 0.95 to 1.75; Figure 3A; Kaplan-Meier estimates).

Figure 3.

Copeptin is associated with combined cardiovascular events and all-cause mortality. Kaplan-Meier estimates for time to combined cardiovascular events (A) and all-cause death (B) in subgroups of patients according to quartiles ofcopeptin.

Finally, death risk was nearly 50% higher in patients with copeptin concentrations within the fourth quartile compared with the first quartile (HR_model1, 1.48; 95% CI, 1.15 to 1.90; HR_model2, 1.24; 95% CI, 0.95 to 1.62; Figure 3B; Kaplan-Meier estimates).

Complete results are shown in Tables 2 and 3.

Table 2.

Risk of stroke, sudden death, MI, death caused by CHF, combined CVE, and all-cause mortality by quartiles of baseline copeptin

Outcome	Copeptin (pmol/L)
	Quartile 1 (≤51.2) (n = 315)	Quartile 2 (>51.2 to (≤80.8) (n = 311)	Quartile 3 (>80.8 to (≤122) (n = 311)	Quartile 4 (>122) (n = 304)
Stroke
number of events (n)	11	28	26	37
Kaplan-Meier estimatea (95% CI)	0.05 (0.02 to 0.08)	0.12 (0.07 to 0.16)	0.10 (0.06 to 0.15)	0.13 (0.09 to 0.18)
adjusted HR (95% CI)	d	2.69 (1.31 to 5.50)	2.34 (1.14 to 4.80)	3.48 (1.71 to 7.09)
adjusted HRb (95% CI)	d	2.40 (1.17 to 4.94)	2.08 (1.01 to 4.31)	3.01 (1.44 to 6.33)
Sudden death
number of events (n)	23	36	54	45
Kaplan-Meier estimatea (95% CI)	0.09 (0.05–0.13)	0.16 (0.11 to 0.21)	0.22 (0.16 to 0.27)	0.18 (0.12 to 0.23)
adjusted HR (95% CI)	d	1.38 (0.81 to 2.36)	2.15 (1.30 to 3.55)	1.73 (1.01 to 2.95)
adjusted HRb (95% CI)	d	1.29 (0.75 to 2.20)	1.92 (1.15 to 3.19)	1.47 (0.84 to 2.57)
MI
number of events (n)	43	50	53	50
Kaplan-Meier estimatea (95% CI)	0.18 (0.13 to 0.24)	0.20 (0.15 to 0.26)	0.25 (0.18 to 0.31)	0.23 (0.17 to 0.29)
adjusted HR (95% CI)	d	0.98 (0.65 to 1.49)	1.14 (0.76 to 1.72)	0.96 (0.62 to 1.50)
adjusted HRb (95% CI)	d	0.95 (0.62 to 1.45)	1.12 (0.73 to 1.71)	0.95 (0.60 to 1.51)
Death caused by CHF
number of events (n)	9	8	8	15
Kaplan-Meier estimatea (95% CI)	0.03 (0.01 to 0.05)	0.03 (0.01 to 0.06)	0.03 (0.01 to 0.06)	0.07 (0.03 to 0.11)
adjusted HR (95% CI)	d	0.72 (0.27 to 1.92)	0.68 (0.25 to 1.82)	1.09 (0.43 to 2.77)
adjusted HRb (95% CI)	d	0.70 (0.26 to 1.86)	0.62 (0.23 to 1.72)	1.09 (0.41 to 2.89)
Combined CVEc
number of events (n)	84	113	127	138
Kaplan-Meier estimatea (95% CI)	0.32 (0.26 to 0.39)	0.42 (0.35 to 0.48)	0.46 (0.40 to 0.53)	0.50 (0.43 to 0.56)
adjusted HR (95% CI)	d	1.19 (0.89 to 1.59)	1.38 (1.04 to 1.83)	1.42 (1.06 to 1.90)
adjusted HRb (95% CI)	d	1.11 (0.83 to 1.49)	1.28 (0.96 to 1.70)	1.29 (0.95 to 1.75)
All-cause death
number of events (n)	113	142	163	191
Kaplan-Meier estimatea (95% CI)	0.42 (0.36 to 0.49)	0.48 (0.41 to 0.54)	0.52 (0.46 to 0.58)	0.56 (0.50 to 0.62)
adjusted HR (95% CI)	d	1.10 (0.85 to 1.42)	1.35 (1.05 to 1.73)	1.48 (1.15 to 1.90)
adjusted HRb (95% CI)	d	1.00 (0.77 to 1.29)	1.19 (0.93 to 1.53)	1.24 (0.95 to 1.62)

^aUnadjusted Kaplan-Meier estimates at the end of year 4.

^bFurther adjustment for markers of volume status and residual urine output.

^cMyocardial infarction, cardiac death, and stroke.

^dPatients with copeptin levels within the first quartile served as reference.

Table 3.

Risk of stroke, sudden death, MI, death caused by CHF, combined CVEs, and all-cause mortality per unit increase in copeptin (continuous variable, log-transformed)

Outcome	HR and 95% CI
	Crude	Adjusted	Adjusteda
Stroke	1.73	1.74	1.62
	(1.25 to 2.39)	(1.23 to 2.47)	(1.11 to 2.36)
	P < 0.001	P = 0.002	P = 0.012
Sudden death	1.49	1.44	1.33
	(1.16 to 1.92)	(1.10 to 1.91)	(0.99 to 1.80)
	P = 0.002	P = 0.009	P = 0.059
MI	1.05	1.00	0.99
	(0.85 to 1.30)	(0.79 to 1.25)	(0.77 to 1.26)
	P = 0.627	P = 0.964	P = 0.90
Death caused by CHF	1.53	1.33	1.37
	(0.93 to 2.53)	(0.75 to 2.36)	(0.74 to 2.55)
	P = 0.098	P = 0.332	P = 0.323
Combined CVEb	1.32	1.24	1.18
	(1.14 to 1.52)	(1.06 to 1.45)	(1.00 to 1.40)
	P < 0.001	P = 0.008	P = 0.052
All-cause death	1.30	1.29	1.18
	(1.15 to 1.50)	(1.12 to 1.48)	(1.02 to 1.36)
	<0.001	P < 0.001	P = 0.032

^aFurther adjustment for markers of volume status and residual urine output.

^bMI, cardiac death, and stroke.

DISCUSSION

This is the first study to highlight the role of copeptin as a risk marker for long-term outcomes in hemodialysis patients. We showed a strong and independent association of plasma copeptin levels with stroke in hemodialysis patients with T2DM. Furthermore, patients with increased copeptin levels show an increased risk for CVE, sudden death, and all-cause mortality, independently of common known risk factors. In contrast, no association between future MI, death caused by CHF, and copeptin was observed.

The prognostic value of copeptin has already been shown in patients with heart failure,^9,10 coronary artery disease,^11,12 and with acute stroke.¹³ However, regarding cerebrovascular events, copeptin has only been evaluated focusing on outcome after stroke. At this time, the predictive value of copeptin in stable patients regarding future stroke is unknown.¹³ Stroke is one of the leading causes of morbidity and mortality worldwide.¹⁴ In prevalent hemodialysis patients 75 to 84 years of age, the probability of death after a stroke is 60% within 1 year compared with 20% in those without a stroke.⁴ Therefore, knowledge of prognostic markers is highly important. In our analysis, baseline copeptin was a strong predictor of stroke in maintenance hemodialysis patients, independently of known risk factors such as arrhythmia, history of stroke, older age,¹⁵ NT-pro BNP,¹⁶ or C-reactive protein.¹⁷ The underlying mechanisms are unknown, but AVP may play a role in the development of cerebral vasospasm and in ischemic brain edema. It exerts potent vasoconstriction by stimulating V1 receptors on smooth muscle cells¹⁸ and inducing secretion of endothelin 1 and prostaglandin D2 from endothelial cells.¹⁹ In addition, AVP seems to be involved in brain edema formation by stimulation of the V1a receptors and influencing blood–brain barrier sodium transporters.^20,21 V1a receptor activation might also involve platelet activation,²² thereby additionally increasing the risk of stroke.

Sudden cardiac death accounts for 26% of all-cause mortality and 60% of cardiovascular deaths in patients with T2DM on hemodialysis.^4,8 Therefore, its prevention remains a major challenge. A number of different causes may account for sudden death in dialysis patients, including micro- and macrovascular disease, sympathetic overactivity, structural heart disease, electrolyte and volume shifts caused by the hemodialysis procedure, and several prognostic markers.^16,17,23,24 This study is the first to show a significant association between sudden death and copeptin.

High concentrations of plasma AVP, as we observed here on the basis of its surrogate, copeptin, are known to preferably stimulate V1a receptors. Therefore, V1a-dependent vasoconstriction that increases afterload, ventricular stress,²⁵ and cardiac hypertrophy²⁶ may result in systemic vascular resistance and a decrease in cardiac output and contractility.²⁷ Several investigators have shown a V1a receptor–mediated coronary vasoconstrictor response to vasopressin,²⁸ which is an effect that seems to be dose dependent. The activation of V1b receptors may play an additional role. Cardiomyocyte mineralocorticoid receptors are normally occupied by cortisol, which acts in a tonic inhibitory fashion. However, in the context of tissue damage and generation of reactive oxygen species, this inhibitory role is transformed, and cortisol becomes a mineralocorticoid agonist, mimicking the effects of aldosterone.^29,30 In line with this, cortisol is associated with mortality in chronic heart failure patients,³¹ and antagonists of the mineralocorticoid receptor lower the incidence of sudden cardiac death in patients with heart failure.^32,33

Although in patients with CHF, copeptin has been described to be a predictor of all-cause mortality and worsening of heart failure,^2,34,35 no association was found between copeptin and the risk of death caused by CHF in this analysis. Most likely, this is explained by the low incidence of events (n = 40). Furthermore, no association was detected between copeptin and future MI. A link between copeptin and myocardial ischemia was proposed in studies of patients with coronary artery disease¹¹ and patients with CHF⁹ after MI.¹⁰ However, until now, copeptin has not been described as a risk factor for future MI in stable patients.

Because the combined CVE consists of stroke, cardiac death (60% are classified as sudden cardiac death), and MI, the association between copeptin and CVE can be interpreted as a reflection of the findings described above. In line with this, copeptin is found to be a risk marker for all-cause mortality in this cohort, which is consistent with findings in patients with heart failure, acute MI, acute stroke, and diabetes mellitus.^{1,13,34,36,37} Whereas these studies investigated shorter-term mortality (60 days to 33 months), median follow-up in the 4D study was 4 years. In healthy individuals, AVP is known to play a central role in fluid and sodium homeostasis, maintaining normal BP and volume status.

Compared with healthy subjects, however, copeptin levels in this study are highly increased (median, 81 pmol/L). In chronic kidney disease, few data exist evaluating copeptin concentrations.^38,39,40 These data found copeptin to be associated with renal function. Reasons for this phenomenon could be that copeptin causes renal function decline, that subjects with low renal function are less sensitive to the actions of copeptin, or that increased copeptin may result from reduced renal elimination. Accordingly, we found copeptin to be correlated with markers of residual renal function, e.g., dialysis vintage and diuretic treatment.

In addition, copeptin is correlated with ultrafiltration volume, suggesting that volume depletion during dialysis might be responsible for an increase in AVP concentrations also. Furthermore, resistance of V2 receptors is present already at earlier stages of chronic kidney disease,^5–7 probably involving activation of feedback mechanisms with a regulatory increase of plasma AVP levels. In addition, left ventricular dysfunction,¹⁰ endothelial stress, and T2DM might also be contributive factors.⁴¹ Actually, increased AVP concentrations in renal failure have already been reported and are associated with biologic activity.⁴² These findings suggest that high copeptin levels may also reflect AVP action in ESRD.

Our study has limitations. It is a post hoc analysis among hemodialysis patients with T2DM, and therefore, the relationship between copeptin and risk may not be generalizable. Furthermore, causality cannot be inferred from observed associations. These associations remained significant in multivariate analyses, suggesting that they are independent of possible confounding factors. However, residual confounding cannot be excluded.

However, what makes copeptin particularly attractive for research is not only its prognostic information but also a putative role of AVP blockade as an intervention strategy. Currently, several vasopressin receptor antagonists are being developed for clinical use.⁴³ Use of these substances may provide crucial information of whether increased AVP secretion is prognostic or causative for cardiovascular disease.

We conclude that copeptin is a potent risk marker for CVE and all-cause death in hemodialysis patients with T2DM. This is mainly because of the strong associations with stroke and sudden death, whereas copeptin does not prove to be a risk marker for MI or death caused by CHF. Whether antagonism of V1 rather than V2 receptors might improve outcome needs to be evaluated in future studies.

CONCISE METHODS

Study Design and Participants

Design and methods of the German Diabetes and Dialysis study have previously been reported in detail.⁸ Briefly, the 4D study was a prospective, randomized trial among 1255 hemodialysis patients with T2DM, 18 to 80 years of age. Between March 1998 and October 2002, patients were recruited in 178 German dialysis centers and randomly assigned to double-blind treatment with 20 mg atorvastatin (n = 619) or placebo (n = 636) once daily. At each follow-up until March 2004, blood samples were taken, and clinical information including any adverse events and an electrocardiogram was recorded. The 4D study was approved by an institutional review committee, and all patients gave written informed consent.

Outcome Measures

The primary endpoint of the 4D study was a composite of cardiac death, nonfatal MI, and stroke, whichever occurred first (CVE). For this analysis, stroke, MI, sudden death, death caused by CHF, CVE, and all-cause mortality were chosen as outcome measures and centrally adjudicated according to prespecified criteria by a specialized committee, blinded to study treatment. Stroke was defined as a neurologic deficit lasting >24 hours. Computed tomography or magnetic resonance imaging was available in all but 16 cases. MI was diagnosed when at least two of three criteria were met: typical symptoms; elevated levels of cardiac enzymes; diagnostic changes in the electrocardiogram. An electrocardiogram documenting silent MI was also considered evidence of MI. Electrocardiograms were evaluated by an independent committee⁴⁴ according to Minnesota classification system. Death from cardiac causes comprised fatal MI, sudden death, death caused by CHF, death during or within 28 days after percutanous coronary intervention or MI, and all other deaths ascribed to coronary artery disease (CAD). Sudden death was considered as death verified by terminal rhythm disorders in an electrocardiogram, observed death within 1 hour after onset of cardiac symptoms, unexpected death, death confirmed by autopsy as being presumably or possibly of cardiac origin, and death in the absence of a potassium level ≥7.5 mmol/L before the start of the three most recent hemodialysis sessions.⁸

Data Collection

Information on patients' demographics, history, comorbidities, and dialysis treatment was obtained through patient interviews and the reports of the respective nephrologists.

CAD was defined by a history of MI, coronary artery bypass grafting surgery, percutaneous coronary intervention, and the presence of CAD as documented by angiography.

BP was measured in the sitting position. Height, weight, and body mass index were measured before dialysis sessions. Weight was further measured after dialysis sessions, and the two measurements were used to calculate the ultrafiltration volume.

Assays

Baseline serum copeptin concentrations were measured in samples taken within 4 weeks before randomization and stored at −80°C. Blood was taken before the start of the dialysis session and before administration of drugs. It was immediately centrifuged and frozen in aliquots. Shipment to the core laboratory was performed on dry ice. Copeptin was measured in-house in a single batch with a commercial sandwich immunoluminometric assay (B.R.A.H.M.S CT-proAVP LIA; B.R.A.H.M.S AG, Henningsdorf/Berlin, Germany). The assay used two polyclonal antibodies to the amino acid sequences 132 to 164 of preprovasopressin in the C-terminal region of the precursor. One antibody is bound to polystyrene tubes, and the other is labeled with acridinium ester for chemiluminescence detection, enabling great precision and dynamics.³ Although there are sound data underlining the excellent stability of the copeptin peptide during short-term storage and four cycles of freezing and thawing, no long-term storage data are available thus far. The lower detection limit was 0.4 pmol/L, median copeptin in 200 healthy individuals was 3.7 pmol/L, and the 97.5 percentile was 16.4 pmol/L. The interassay coefficient of variation was 6.0%.

Statistical Analysis

Patient characteristics are presented according to quartiles of baseline copeptin. Spearman correlation coefficients and scatter plots with linear quadratic regression equations were calculated to estimate the association between copeptin and further variables.

The effect of copeptin on outcome was assessed by Kaplan-Meier estimates for incidences of the prespecified endpoints grouped by copeptin quartiles and by relative risks derived from Cox regression analyses, i.e., HRs and corresponding 95% CIs. First, analyses were adjusted for gender, age, smoking, BP, body mass index, LDL cholesterol, C-reactive protein, history of stroke or transitory ischemic attack, CAD, peripheral vascular disease, CHF, arrhythmia, and arterio-venous fistula, phosphate, hemoglobin, glycated hemoglobin, albumin, and atorvastatin treatment (model 1). Second, analyses were further adjusted for markers of volume status and residual urine output (ultrafiltration volume, Nt-pro-BNP, dialysis vintage, and diuretic use at baseline) (model 2).

Cox regression analyses included copeptin as categorical (baseline copeptin quartiles) and as continuous variables (i.e., logarithmically transformed, because values were not normally distributed and the linearity assumption was satisfied for the log-transformed values).

Analyses were done using SAS version 9.1.3.

DISCLOSURES

None.