2288 Diabetes Care Volume 39, December 2016

Gilberto Velho,1 Ray El Boustany,1 Plasma Copeptin, Kidney Guillaume Lefevre,` 2 Kamel Mohammedi,1,3 Outcomes, Ischemic Heart Disease, Fred´ eric´ Fumeron,1,4 Louis Potier,1,3,4 Lise Bankir,1,5 Nadine Bouby,1,5 and All-Cause Mortality in People Samy Hadjadj,6,7,8,9 Michel Marre,1,3,4 and With Long-standing Type 1 Ronan Roussel1,3,4 Diabetes Diabetes Care 2016;39:2288–2295 | DOI: 10.2337/dc16-1003

OBJECTIVE Plasma copeptin, a surrogate for vasopressin, has been associated with a decline in renal function and albuminuria in population-based studies as well as with progression of diabetic nephropathy in people with type 2 diabetes. We assessed the risk of kidney and coronary events and all-cause mortality associated with plasma copeptin in people with type 1 diabetes.

RESEARCH DESIGN AND METHODS Plasma copeptin was measured in baseline samples of the GENEDIAB (n = 398; 56% 1INSERM, UMR_S 1138, Centre de Recherche des male; mean 6 SD age 45 6 12 years and diabetes duration 28 6 10 years) and Cordeliers, Paris, France 2 GENESIS (n = 588; 52% male; age 42 6 11 years; diabetes duration 27 6 9years) Assistance Publique Hopitauxˆ de Paris, Hopitauxˆ Universitaires Est Parisien–Tenon, Ser- cohorts. Follow-up data were available for 218 GENEDIAB and 518 GENESIS par- vice de Biochimie et Hormonologie, Paris, France ticipants. Median duration of follow-up was 10.2 and 5.0 years, respectively. 3Assistance Publique Hopitauxˆ de Paris, Hopitalˆ Bichat, DHU FIRE, Department´ de Diabetologie,´ RESULTS Endocrinologie et Nutrition, Paris, France 4 Upper sex-specific tertiles of copeptin were associated with a higher incidence of Universite´ Paris Diderot, Sorbonne Paris Cite,´ UFR de Medecine,´ Paris, France end-stage renal disease (ESRD) during follow-up (hazard ratio [HR] for third vs. 5Sorbonne Universites,´ Universite´ Pierre et Marie first tertile 26.5 [95% CI 8.0–163.3; P < 0.0001]; analysis in pooled cohorts adjusted Curie-Paris 6, Paris, France for age, sex, duration of diabetes, and cohort membership). The highest tertile of 6Department´ de Endocrinologie et Diabetologie,´ PATHOPHYSIOLOGY/COMPLICATIONS CHU de Poitiers, Poitiers, France copeptin was also associated with incidence of myocardial infarction or coronary 7 – P INSERM, Unite´ de Recherche 1082, Poitiers, revascularization (HR 2.2 [95% CI 1.2 4.0]; = 0.01) and all-cause mortality (HR 3.3 France [95% CI 1.8–6.5]; P < 0.0001) during follow-up. 8INSERM, CIC 1402, Poitiers, France 9Universite´ de Poitiers, UFR de Medecine´ et Phar- CONCLUSIONS macie, Poitiers, France Plasma copeptin is a predictor for the risk of ESRD, coronary heart disease, and all- Corresponding author: Gilberto Velho, gilberto. cause mortality in people with type 1 diabetes. Results are consistent with data [email protected]. from experimental and epidemiological studies, suggesting that high circulating Received 8 May 2016 and accepted 17 Septem- levels of vasopressin are deleterious to renal function. ber 2016. This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/ Circulating levels of vasopressin (or antidiuretic hormone) are increased in people suppl/doi:10.2337/dc16-1003/-/DC1. with type 1 or type 2 diabetes and in animal models with spontaneous or strepto- © 2016 by the American Diabetes Association. zotocin-induced diabetes (1). Vasopressin is cosecreted into the blood by the Readers may use this article as long as the work is properly cited, the use is educational and not neurohypophysis in an equimolar amount with copeptin, the COOH-terminal por- for profit, and the work is not altered. More infor- tion of the preprovasopressin peptide. The main stimuli for the secretion of vaso- mation is available at http://www.diabetesjournals pressin are an increase in plasma osmolality and/or a decrease in arterial circulating .org/content/license. care.diabetesjournals.org Velho and Associates 2289

volume. The causes of increased vaso- RESEARCH DESIGN AND METHODS 20–200 mg/min or UAC = 20–200 mg/L] pressin and copeptin levels in diabetes Participants and plasma creatinine ,150 mmol/L), 3) are not fully elucidated. These high lev- Gen´ etique´ de la Nephropathie´ Diabetique´ established nephropathy (past or present els do not seem to result from an in- (GENEDIAB) and Genesis France-Belgium macroalbuminuria [UAE .300 mg/24 h crease in plasma osmolality due to (GENESIS) are multicenter binational co- or .200 mg/min or UAC .200 mg/L] hyperglycemia but could result from a horts of people with long-standing type 1 and plasma creatinine ,150 mmol/L), relative contraction of extracellular vol- diabetes that studied the vascular com- and 4) advanced nephropathy (past or ume induced by glycosuria and/or from plications of diabetes. GENEDIAB in- present macroalbuminuria and plasma an increased sensitivity of hypothalamic cluded 494 participants selected on the creatinine .150 mmol/L or history of osmoreceptor neurons to the plasma basis of a diagnosis of type 1 diabetes end-stage renal disease [ESRD] defined osmoticload(2).Fromanadaptiveper- before the age of 35 years and past or as hemodialysis requirement or kidney spective, high levels of vasopressin may present diagnosis of severe diabetic ret- transplantation). Diabetic retinopathy be beneficial in the short term by limit- inopathy (19). GENESIS was a family- was staged according to Kohner’sclassifi- ing the water loss in urine induced by based study conducted in 578 first-degree cation as nonproliferative, preprolifera- the higher osmolar load due to glycos- relatives and 662 probands with type 1 tive, or proliferative (21) based on direct uria (1). However, in the long term, per- diabetes selected on the basis of a diag- funduscopy and/or fluorescein angiogra- sistently high levels of vasopressin nosis of diabetes before the age of phy results. Systolic and diastolic blood might aggravate hyperglycemia by in- 35 years and past or present diagnosis pressure were measured in supine posi- creasing hepatic glucose production of diabetic retinopathy (20). Subsets of tion with a mercury sphygmomanometer and inducing insulin resistance (3) and participants from GENEDIAB (n = 260) and are reported as the mean of two con- might be deleterious to renal function and GENESIS (n = 550) were included secutive measurements (19). Myocardial (4). Because vasopressin secretion can in a prospective observational study and infarction was diagnosed on the basis of be modulated by water intake (5) and followed until an end point was reached at least two of the following three criteria: its actions by nonpeptide selective or until February 2007. The subsets com- constrictive chest pain lasting $20 min, receptor antagonists (vaptans) (6), the prised participants who attended outpa- increased serum creatine phosphokinase vasopressin/hydration system could tient clinics at least once during the activity and/or troponin concentration, or be a potential therapeutic target for follow-up period. Median (interquartile typical electrocardiographic changes. the prevention and treatment of dia- range) duration of follow-up was 10.2 Cause of death was established from hos- betic complications, notably diabetic (2.7) and 5.0 (1.6) years for GENEDIAB pital records. Missing data were obtained nephropathy (4). and GENESIS, respectively. In the present by a phone interview with the patient’s Circulating levels of vasopressin and investigation, copeptin was effectively general practitioner and/or by consulting copeptin correlate over a wide range of measured in 398 GENEDIAB participants the death certificate national registry. We plasma osmolality (7,8), and copeptin, and 588 GENESIS probands for whom considered the incidence of three out- whichiseasiertoassay,isanadequate plasma samples were available at base- comes during follow-up: ESRD, a coronary surrogate marker of vasopressin and line, including 218 GENEDIAB and 518 event (defined as the occurrence of myo- hydration status (7,8). Plasma copeptin GENESIS participants seen at follow-up. cardial infarction and/or coronary revas- is associated with hyperglycemia and The flowchart of participants is shown in cularization procedure), and all-cause type 2 diabetes as well as with other Supplementary Fig. 1. Study protocols mortality. Characteristics of GENEDIAB cardiovascular risk factors, such as were approved by the ethics committee and GENESIS participants at baseline by arterial hypertension, abdominal obe- of the Angers University Hospital (Angers, the incidence of ESRD and all-cause mor- sity, insulin resistance, and the meta- France), and all participants gave written tality during follow-up were published bolic syndrome (9–14). Plasma copeptin informed consent. previously (22,23). Incidence rates and also has been associated with the de- clinical data reported in the present study cline in renal function and albuminuria Definition of Clinical Parameters and are those of the subsets of participants in population-based studies (4,15,16) Outcomes with an available plasma sample at base- and with the development and pro- An ad hoc event committee reviewed line for copeptin assay. gression of diabetic nephropathy in the case record of each patient to vali- people with type 2 diabetes (17,18). date baseline data (19) and later and the Laboratory Procedure However, prospective data are lacking incidence of outcomes during follow-up. Copeptin concentration was measured in people with type 1 diabetes. We Diabetic nephropathy stages were de- in 2014 in fasting plasma-EDTA samples assessed the association of plasma fined according to the following criteria: collected at baseline and kept frozen at copeptin at baseline with the risk of 1) no nephropathy (urinary albumin ex- 280°C. An automated immunoluminomet- subsequent kidney and coronary mor- cretion [UAE] ,30 mg/24 h or ,20 mg/min ric assay (ultra-sensitive B.R.A.H.M.S bidity and all-cause mortality in people or urinary albumin concentration [UAC] Copeptin proAVP; Thermo Scientific, Hen- with long-standing type 1 diabetes. We ,20 mg/L and plasma creatinine nigsdorf, Germany) on a KRYPTOR com- tested the hypothesis that the associa- ,150 mmol/Linatleasttwoofthree pact PLUS system was used (24). The limit tions with coronary morbidity and all- consecutive assessments and in the ab- of detection was 0.9 pmol/L. The intra- cause mortality could reflect the kidney sence of antihypertension treatment), 2) assay coefficient of variance reported by morbidity associated with these incipient nephropathy (persistent micro- the manufacturer was ,15% and ,8% outcomes. albuminuria [UAE = 30–300 mg/24 h or for concentrations ranging from 2.0 to 2290 Copeptin and Vascular Complications in Diabetes Diabetes Care Volume 39, December 2016

4.0 pmol/L and 4.0 to 15.0 pmol/L, re- groups were assessed by Pearson x2 residuals (estat phtest, Stata 12.1 soft- spectively. The reported interassay coef- test, Wilcoxon and Kruskal-Wallis test, ware; StataCorp, College Station, TX). ficient of variance was ,18% and ,10%, ANOVA, or ANCOVA. Association of Kaplan-Meier curves were used to plot respectively, for the lower and higher co- plasma copeptin with diabetic nephrop- the incidence of outcomes over time. peptin concentration range. athy at baseline was assessed by logistic To increase statistical power, all pro- regression analyses. Cox proportional spective analyses were performed in Computations and Statistical Analyses hazards regression models were used GENEDIAB and GENESIS pooled co- Tertiles of plasma copeptin concentra- to examine the effect of plasma copep- horts. Data were log-transformed for tion were computed separately for tin at baseline on time-related survival the analyses when the normality of womenandmentotakeintoaccount (outcome-free) rates in prospective the distribution was rejected by the the well-known sex-related differences analyses. Adjustments for clinical and Kolmogorov-Smirnoff-Lillifors goodness- on copeptin levels (8,10–14,25). Esti- biological parameters were carried of-fit test. Statistical analyses were mated glomerular filtration rate (eGFR) out by including these parameters as performedwithJMP(SASInstitute, was calculated by the MDRD equation covariates in the regression models. Cary, NC) and Stata 12.1 software. (26). The annual variation of eGFR dur- Odds ratios (ORs) or hazard ratios P , 0.05 was considered significant. ing the study was computed as the dif- (HRs) with their 95% CIs were com- ference between values at the end of puted in these analyses, respectively. RESULTS follow-up and at baseline divided by For all outcomes in the Cox models, Clinical Characteristics at Baseline the duration of follow-up. Results are the validity of proportional hazards as- Participant characteristics at baseline by expressed as mean 6 SD, except where sumption was verified for all covari- tertiles of plasma copeptin are shown in stated otherwise. Differences between ates by a test based on Schoenfeld Table 1. Participants in the highest

Table 1—Clinical characteristics at baseline by tertiles of plasma copeptin GENEDIAB study GENESIS study T1 T2 T3 P value T1 T2 T3 P value n 135 132 131 196 195 197 Age (years) 46 6 12 44 6 13 44 6 12 0.21 43 6 10 41 6 11 42 6 11 0.61 Plasma copeptin (pmol/L) Women 1.9 (0.9–2.9) 4.5* (3.0–7.9) 14.6*† (8.0–49.8) ,0.0001 2.2 (0.9–2.8) 3.9* (2.8–5.9) 13.8*† (6.1–153.1) ,0.0001 Men 2.6 (0.9–3.9) 5.9* (4.0–9.1) 19.0*† (9.3–158) ,0.0001 3.3 (0.9–4.7) 6.2* (5.0–8.7) 16.3*† (8.8–210.6) ,0.0001 Age at diagnosis of diabetes (years) 18 6 9156 9* 16 6 90.04156 8156 10 15 6 10 0.99 Duration of diabetes (years) 28 6 10 28 6 10 28 6 90.97276 9266 9276 10 0.46 BMI (kg/m2) 23.7 6 3.2 24.0 6 3.1 23.6 6 3.3 0.64 24.7 6 3.7 24.3 6 3.5 24.0 6 3.6 0.20

HbA1c 0.10 0.001 %8.46 1.8 8.9 6 1.9 8.6 6 1.8 8.3 6 1.1 8.5 6 1.3 9.0 6 1.5* mmol/mol 68 6 20 73 6 20 70 6 20 67 6 12 70 6 14 72 6 17 Systolic BP (mmHg) 134 6 18 137 6 18 143 6 19* 0.0004 129 6 16 131 6 19 136 6 22* 0.01 Diastolic BP (mmHg) 77 6 13 79 6 10 81 6 12* 0.01 74 6 8756 10 78 6 11*† 0.002 Antihypertensive treatment (%) 46.7 48.0 69.8 0.0001 36.7 46.2 70.4 ,0.0001 eGFR (mL/min/1.73 m2) 79 (26) 78 (35) 49 (41)*† ,0.0001 91 (36) 91 (43) 58 (59)*† ,0.0001 UAC (mg/L) 8 (35) 26 (154)* 425 (1,398)*† ,0.0001 10 (28) 13 (62)* 298 (1,183)*† ,0.0001 Diabetic nephropathy stage (%) ,0.0001 ,0.0001 Absent 48 40 14 69 59 20 Incipient 27 27 11 20 23 18 Established 18 24 24 9 12 26 Advanced 7 9 51 2 6 36 Use of diuretics (%) 10.1 13.2 33.3 ,0.0001 5.1 7.2 32.7 ,0.0001 Use of ACE-I or ARB (%) 37.8 41.3 50.0 0.13 28.6 36.4 61.7 ,0.0001 Previous MI (%) 7.4 7.6 8.4 0.94 4.1 4.6 5.6 0.78 Data are mean 6 SD, median (range) (copeptin), and median (interquartile range) (eGFR and UAC), unless otherwise stated. Statistics for quantitative parameters are ANOVA with log-transformed data or Wilcoxon test (eGFR and UAC). Tukey Kramer honest significant difference test following ANOVA or Wilcoxon test were used for comparisons of each pair: *significantly different (P , 0.05) from T1; †significantly different (P , 0.05) from T2. Arterial hypertension was defined as the use of antihypertensive drugs and history of hypertension. ACE-I, ACE inhibitor; ARB, angiotensin receptor blocker; BP, blood pressure; MI, myocardial infarction. care.diabetesjournals.org Velho and Associates 2291

2 tertile had higher HbA1c levels (GENESIS 0.46 (T3) mL/min/1.73 m per year (ANCOVA of ESRD remained significant in sensi- study only) and systolic and diastolic P = 0.002), adjusted for age, duration of tivity analyses when the cohorts were blood pressures and were more likely diabetes, HbA1c, UAC (normo-, micro-, analyzed separately, whereas the associa- to have a history of arterial hyperten- or macroalbuminuria), eGFR at base- tion with the incidence of coronary events sion. They had lower eGFR and higher line, sex, cohort membership, and remained significant in the GENESIS cohort UAC. Diabetic nephropathy at baseline duration of follow-up. Cumulative inci- only (Supplementary Table 2). was more frequent and more severe dence of ESRD and coronary events Cox proportional hazards regression in participants in the highest tertile during follow-up was 7.0% (n = 47) analyses were also performed in the (T3) of plasma copeptin but was similar and 9.4% (n = 67), respectively. The in- pooled cohorts in a multiadjusted model. in participants in tertiles T1 and T2 cidence rates of theses outcomes were Clinical and biological parameters at base- (Supplementary Fig. 2). Regression anal- 1.16 and 1.55 per 100 person-years, re- line with P # 0.20 at least in one of the yses confirmed the associations of the spectively. Characteristics of GENEDIAB cohorts in the comparison by tertiles of highest tertile of plasma copeptin with and GENESIS participants at baseline plasma copeptin (Table 1) were entered the prevalence of established and ad- by the incidence of outcomes during in the model as independent covariates, vanced diabetic nephropathy at base- follow-up are shown in Supplementary together with cohort membership, sex, line in both cohorts. The ORs (95% CI) Table 1. The incidence of ESRD and cor- age, duration of diabetes, and previous for established and advanced nephrop- onary events during follow-up in the history of myocardial infarction (for athy in GENEDIAB were 3.43 (1.77–6.81; pooled cohorts by tertiles of baseline the Cox of coronary events) at baseline. P = 0.0002) and 24.48 (11.97–53.72; P , plasma copeptin is shown in Table 2 Plasma copeptin remained significantly as- 0.0001) for T3 versus T1 and T2, ad- and Fig. 1. Incidence of each outcome sociated with the incidence of ESRD in the justed for sex, age, and duration of di- was significantly higher for partici- multiadjusted model (Table 2, model 2), abetes. The ORs were 7.91 (4.64–13.68; pants in the upper tertile of plasma althoughwithanattenuatedHRcom- P , 0.0001) and 31.29 (16.64–62.38; copeptin. Cox proportional hazards re- pared with the minimally adjusted model. P , 0.0001), respectively, in GENESIS. gression models confirmed a positive The inclusion of baseline UAC and eGFR in association of the upper tertile of the multiadjusted model was responsible Copeptin and Kidney and Coronary plasma copeptin and loge[copeptin] for most of the HR attenuation. UAC, Outcomes During Follow-up with the incidence of ESRD and coro- eGFR, HbA1c, BMI, diastolic blood pres- The mean 6 SEM annual decline in eGFR nary events in a minimally adjusted sure, and use of diuretic and antihyperten- during follow-up in the GENESIS and model, including as covariates cohort sive medications at baseline also remained GENEDIAB pooled study by tertiles of membership,sex,age,andduration associated with the incidence of ESRD dur- baseline plasma copeptin was 23.09 6 of diabetes at baseline (Table 2, model ing follow-up (Supplementary Fig. 3). Only 0.37 (T1), 23.74 6 0.36 (T2), and 25.18 6 1). Associations with the incidence age, diastolic blood pressure, UAC, and a history of myocardial infarction at baseline remained associated with the incidence of coronary events in the Table 2—Clinical outcomes during GENESIS and GENEDIAB follow-up by plasma copeptin at baseline multiadjusted model (Table 2, model 2, Clinical outcome during follow-up and Supplementary Fig. 3). When UAC, eGFR, and systolic and diastolic blood ESRD Coronary event All-cause mortality pressure were excluded from the multi- No Yes No Yes No Yes adjusted model, copeptin became associ- T1 248 (99.2) 2 (0.8) 238 (92.6) 19 (7.4) 246 (94.6) 14 (5.4) ated with the incidence of coronary T2 228 (95.0) 12 (5.0) 224 (91.4) 21 (8.6) 230 (91.6) 21 (8.4) events (HR for 1 unit of loge[copeptin] T3 149 (81.9) 33 (18.1) 186 (87.3) 27 (12.7) 189 (85.5) 32 (14.5) 1.3 [95% CI 1.0–1.7]; P = 0.04), suggest- ing that these covariates accounted for the HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value effect of copeptin on the outcome. Model 1 T3 vs. T1 26.5 (8.0–163.3) ,0.0001 2.2 (1.2–4.0) 0.01 3.3 (1.8–6.5) ,0.0001 T3 vs. T2 4.4 (2.3–8.8) ,0.0001 1.7 (0.9–3.1) 0.07 2.0 (1.1–3.5) 0.02 Copeptin and Mortality During T2 vs. T1 6.1 (1.7–39.1) 0.005 1.3 (0.7–2.4) 0.48 1.7 (0.9–3.5) 0.13 Follow-up

Loge[copeptin] 4.3 (3.1–5.9) ,0.0001 1.4 (1.1–1.7) 0.009 1.7 (1.4–2.1) ,0.0001 The cumulative incidence of all-cause Model 2 mortality during follow-up in the GENESIS T3 vs. T1 9.5 (1.8–174.71) 0.004 1.2 (0.6–2.5) 0.60 1.6 (0.7–3.4) 0.25 and GENEDIAB pooled study was 9.2% T3 vs. T2 1.5 (0.7–3.4) 0.33 1.0 (0.5–2.1) 0.98 1.3 (0.6–2.6) 0.52 (n = 67), and its incidence rate was 1.43 – – – T2 vs. T1 6.4 (1.1 118.4) 0.03 1.2 (0.6 2.4) 0.58 1.2 (0.6 2.5) 0.57 per 100 person-years. Characteristics of Log [copeptin] 1.7 (1.2–2.6) 0.005 1.1 (0.8–1.5) 0.58 1.3 (0.9–1.8) 0.13 e GENEDIAB and GENESIS participants at Data are n (%) unless otherwise indicated. Analyses were performed in pooled cohorts. HRs were baseline by the incidence of all-cause computed by Cox proportional hazards regression models for tertiles of plasma copeptin and for mortality during follow-up are shown 1unitofloge[copeptin]. Model 1: adjusted for cohort membership, sex, age, and duration of diabetes at baseline. Model 2: model 1 plus adjustments for eGFR; UAC; systolic and diastolic blood in the Supplementary Table 1. The in- pressure; BMI; HbA1c; and use of ACE inhibitor or angiotensin receptor blocker, diuretic, and cidence of all-cause mortality during antihypertensive medications at baseline. Model 2 for coronary events and all-cause mortality were follow-up was significantly higher for further adjusted for history of myocardial infarction at baseline. P , 0.05 is significant. participants in the upper tertile of 2292 Copeptin and Vascular Complications in Diabetes Diabetes Care Volume 39, December 2016

mortality in the multiadjusted model (Table 2, model 2, and Supplementary Fig. 3). When UAC and eGFR were excluded from the multiadjusted model, copeptin became associated with all-cause mortality (HR for 1 unit of loge[copeptin] 1.4 [95% CI 1.1–1.8]; P = 0.01), suggesting that these covariates accounted for the effect of copeptin on the outcome. Causes of death during follow-up in- cluded cardiovascular complications (38.8%), infectious diseases (16.4%), cancer (10.4%), acute metabolic com- plications (9.0%), renal failure (3.0%), and other or undetermined etiologies (22.4%). We assessed associations of co- peptin with cause-specific mortality grouped as cardiometabolic mortality (death caused by cardiovascular compli- cations, acute metabolic complications, or renal failure) and all other causes of death. Cox proportional hazards regres- sion models confirmed the association of high copeptin levels at baseline with cardiometabolic mortality (HR 2.9 [95% CI 1.3–7.0; P = 0.009] for T3 vs. T1; HR 1.6 [95% CI 1.2–2.2; P = 0.002] for 1 unit of loge[copeptin] adjusted for cohort membership, sex, age, and duration of diabetes at baseline) and with other causes of death (HR 4.6 [95% CI 1.8– 14.1; P = 0.001] for T3 vs. T1; HR 1.9 [95% CI 1.4–2.5; P = 0.0001] for 1 unit of loge[copeptin], same adjustments as above). In the multiadjusted model, plasma copeptin was not significantly associated with cardiometabolic mortal- ity (HR for 1 unit of loge[copeptin] 1.1 [95% CI 0.7–1.7]; P = 0.74), but remained associated with other causes of death (HR for 1 unit of loge[copeptin] 1.8 [95% CI 1.1–2.8]; P = 0.02). When UAC, eGFR, and systolic and diastolic blood pressure were excluded from the multi- adjusted model, copeptin became asso- ciated with cardiometabolic mortality Figure 1—Kaplan-Meier curves for the cumulative incidence of ESRD, coronary events, and all- (HR for 1 unit of loge[copeptin] 1.5 cause mortality during GENEDIAB and GENESIS follow-up by tertiles of baseline plasma copeptin. – Median duration of follow-up (interquartile range) was 5.5 (4.6), 5.4 (4.7), and 5.5 (3.9) years [95% CI 1.1 2.1]; P = 0.02). (P = 0.28) for T1, T2, and T3, respectively. Data are from the pooled cohorts. CONCLUSIONS In the present study, people with long- plasma copeptin at baseline (Table 2 analyzed separately (Supplementary Table standing type 1 diabetes had high levels and Fig. 1). Cox proportional hazards re- 2). Population attributable risk of the up- of plasma copeptin associated with the gression models confirmed a positive per tertile of plasma copeptin for all- prevalence of established and advanced association of the upper tertile of cause mortality during follow-up in the diabetic nephropathy at baseline and plasma copeptin with all-cause mortality pooled cohorts was 0.27. Cohort member- with an increased risk during follow-up in the minimally adjusted model (Table 2). ship (GENEDIAB), HbA1c,BMI,UAC,anda of ESRD, coronary events, and all-cause The association remained significant in sen- history of myocardial infarction at base- mortality. The association with the sitivity analyses when the cohorts were line remained associated with all-cause incidence of ESRD during follow-up care.diabetesjournals.org Velho and Associates 2293

remained significant after adjustment tolvaptan, a V2-receptor antagonist, The risks for ischemic heart disease for relevant confounding factors at for 36 months was associated with and cardiometabolic mortality in the baseline, such as age, duration of diabe- slowed kidney growth and functional current study were influenced by markers tes, blood pressure, HbA1c,eGFR,and decline and with a reduced frequency of of nephropathy (ie, UAC, eGFR), and UAC. On the other hand, the association autosomal-dominant polycystic kidney theassociationofcopeptinwiththese of high plasma copeptin with increased disease–related complications (30). In outcomes depended on these markers risk of coronary events during follow-up renal transplant recipients, high copep- and on arterial hypertension, mostly could be accounted for by the higher tin level at baseline was associated as a consequence of kidney disease, blood pressure, eGFR, and UAC ob- with a faster decline in GFR during a 3- in people with type 1 diabetes. The re- served in participants with higher to 4-year follow-up (31). Plasma copeptin sults agree with data from the litera- plasma copeptin. Finally, the associa- was associated with renal function de- ture showing that the presence and tion with increased mortality risk was cline in people with newly diagnosed severity of chronic kidney disease is not restricted to death from cardio- type 2 diabetes (17) and with the risk the dominant contributor to the excess metabolic causes but was significant of severe kidney outcomes (doubling mortality associated with type 1 diabetes when we considered other causes of of plasma creatinine concentration (39,40). Impaired kidney function may ag- death, including cancer and infectious and/or ESRD) in people with type 2 di- gravate other cardiovascular risk fac- diseases. The association of copeptin abetes and albuminuria (18). Other tors, such as hypertension, oxidative with other causes of death was inde- studies reported associations with car- stress, insulin resistance, dyslipidemia, pendent from kidney phenotypes, but diovascular disease and/or mortality. inflammation, and arterial calcification. the association with cardiometabolic In the population-based MalmoDiet¨ Thus, the association of copeptin with death could be accounted for by the and Cancer Study–Cardiovascular co- ischemic heart disease and cardiometa- higher UAC and lower eGFR observed hort, plasma copeptin was associated bolic mortality observed in the present in participants in the upper tertile of with a combined end point comprising investigation may be partly accounted plasma copeptin. The population at- coronary heart disease, heart failure, for by the deleterious effects of vaso- tributable risk of the upper tertile of and death in people with diabetes but pressin on kidney function. However, plasma copeptin at baseline (.6.1 not in people without diabetes (32). vasopressin induces platelet aggre- pmol/L for women and .8.0 pmol/L However, in older Swedish subjects, gation and has a vasoconstrictor ef- formen)withregardtodeathfrom copeptin was associated with an in- fect on vascular smooth muscle cells all causes was 27%. This figure corre- creasedriskofcoronaryheartdisease through V1aR (35). To our knowledge, sponds to the reduction in mortality and cardiovascular mortality in people the role these vasopressin effects play that could be observed if all partici- with and without diabetes (33). Finally, in the pathophysiology of ischemic heart pants had plasma copeptin below in a study of patients with type 2 disease has not been evaluated. these levels. diabetes undergoing hemodialysis, There are limitations to this study. To our knowledge, this prospective high plasma copeptin was associ- Because of the observational design, investigation is the first to evaluate co- ated with increased risk for all-cause we could not establish a causal relation- peptin as a risk factor for kidney and mortality (34). ship between copeptin/vasopressin and cardiovascular morbidity and mortality Vasopressin binds to three different the outcomes. Moreover, we used co- specifically in people with type 1 diabe- G-protein–coupled receptors. V1aR peptin as a surrogate of vasopressin. tes. In a recent cross-sectional study in is widely expressed, particularly in Plasma concentrations of the peptides 209 adults with type 1 diabetes, copep- vascular smooth muscle cells, hepa- correlate over a wide range of plasma tin correlated positively with the urinary tocytes, platelets, and the central and/or urine osmolalities (7,8), but albumin-to-creatinine ratio and a coro- nervous system. V1bR is expressed in the ratio between their concentra- nary artery calcium score and negatively the endocrine pancreas, in cells of the tions and the strength of their corre- with eGFR (27). Other studies published anterior pituitary, and throughout the lation varies with the GFR (8). Another so far concerned subjects from the gen- brain. V2R is predominantly expressed limitation was the relatively small eral population or patients with kidney in the kidney collecting ducts and in number of outcomes observed during disease or type 2 diabetes. In the gen- the endothelium. Vasopressin stimu- follow-up, potentially reducing the eral population, plasma copeptin has lates hepatic gluconeogenesis and gly- statistical power to observe indepen- been positively associated with renal cogenolysis through V1aR (3,35) and dent associations. Despite these limi- function decline and progression to the release of adrenocorticotropic tations, the study has major strengths. chronic kidney disease stage 3 or worse hormone and glucagon or insulin (de- It includes a cross-sectional analysis of (16), the prevalence or the incidence pending on the extracellular glucose ;1,000 patients with long-standing type 1 of microalbuminuria (12,15,25), and level) through V1bR (35,36). Experi- diabetes from two independent cohorts the presence of renal cysts (15) and in- mental evidence strongly supports a plus a longitudinal study over 5–10 years versely associated with kidney length causal and direct role of vasopressin for .700 participants with comprehensive (15). In patients with autosomal-dominant in the development and aggravation renal, cardiovascular, and survival out- polycystic kidney disease, copeptin was of chronic kidney disease through comes, including cause-specific mortality associated with markers of disease se- V2R activation (37,38), but the mecha- data. verity, including GFR and albuminuria nisms of these deleterious effects are In conclusion, this study in people (28,29). In addition, administration of not fully understood (4). with long-standing type 1 diabetes, 2294 Copeptin and Vascular Complications in Diabetes Diabetes Care Volume 39, December 2016

together with a growing body of data 6. Decaux G, Soupart A, Vassart G. Non-peptide patients with retinopathy or nephropathy: the from experimental and epidemiological arginine-vasopressin antagonists: the vaptans. Genesis France-Belgium Study. Diabetes Care – – studies, suggests that high levels of cir- Lancet 2008;371:1624 1632 2004;27:2661 2668 7. Balanescu S, Kopp P, Gaskill MB, Morgenthaler 21. Kohner EM. The lesions and natural history culating vasopressin are deleterious to NG, Schindler C, Rutishauser J. Correlation of of diabetic retinopathy. In Textbook of Diabetes. kidney function. Intervention studies plasma copeptin and vasopressin concentrations Pickup J, Williams G, Eds. Oxford, Blackwell Sci- are required to assess the potential ben- in hypo-, iso-, and hyperosmolar states. J Clin entific, 1991, p. 575–588 efit of reducing vasopressin secretion or Endocrinol Metab 2011;96:1046–1052 22. Patente TA, Mohammedi K, Bellili-Munoz~ action in the prevention of kidney dis- 8. Roussel R, Fezeu L, Marre M, et al. Compar- N, et al. Allelic variations in the CYBA gene of fi ison between copeptin and vasopressin in a NADPH oxidase and risk of kidney complications ease. The study con rms in people with population from the community and in people in patients with type 1 diabetes. Free Radic Biol type 1 diabetes that copeptin is a risk with chronic kidney disease. J Clin Endocrinol Med 2015;86:16–24 marker for severe clinical outcomes, in- Metab 2014;99:4656–4663 23. Mohammedi K, Bellili-Munoz~ N, Marklund cluding ESRD, ischemic heart disease, 9. Saleem U, Khaleghi M, Morgenthaler NG, SL, et al. Plasma extracellular superoxide dismu- and death. et al. Plasma carboxy-terminal provasopressin tase concentration, allelic variations in the (copeptin): a novel marker of insulin resistance SOD3 gene and risk of myocardial infarction and metabolic syndrome. J Clin Endocrinol and all-cause mortality in people with type 1 Metab 2009;94:2558–2564 and type 2 diabetes. Cardiovasc Diabetol 2015; Acknowledgments. The authors acknowledge 10. Enhorning¨ S, Wang TJ, Nilsson PM, et al. 14:845 Na¨ıma Bellili-Munoz~ (INSERM UMR_S 1138, Plasma copeptin and the risk of diabetes melli- 24. Fenske W, Stork¨ S, Blechschmidt A, Maier Paris, France) for assistance with managing the tus. Circulation 2010;121:2102–2108 SG, Morgenthaler NG, Allolio B. Copeptin in the data bank and plasma samples for copeptin 11. Abbasi A, Corpeleijn E, Meijer E, et al. Sex differential diagnosis of hyponatremia. J Clin measurement. differences in the association between plasma Endocrinol Metab 2009;94:123–129 Funding. R.E.B. was supported by a CIFRE copeptin and incident type 2 diabetes: the Pre- 25. Meijer E, Bakker SJ, Halbesma N, de Jong PE, (Conventions Industrielles de Formation par la vention of Renal and Vascular Endstage Disease Struck J, Gansevoort RT. Copeptin, a surrogate Recherche) grant from INSERM and Danone (PREVEND) study. Diabetologia 2012;55:1963– marker of vasopressin, is associated with micro- Research Centre for Specialised Nutrition. 1970 albuminuria in a large population cohort. Kidney This work was supported by grants from the 12. Enhorning S, Bankir L, Bouby N, et al. Int 2010;77:29–36 Association Diabete` et Risque Vasculaire, DHU Copeptin, a marker of vasopressin, in abdominal 26. Levey AS, Bosch JP, Lewis JB, Greene T, FIRE (Fibrosis, Inflammation, Remodeling in obesity, diabetes and microalbuminuria: the Rogers N, Roth D; Modification of Diet in Renal Cardiovascular, Respiratory and Renal Dis- prospective Malmo¨ Diet and Cancer Study Disease Study Group. A more accurate method eases), and Societ´ e´ Francophone du Diabete.` cardiovascular cohort. Int J Obes 2013;37:598– to estimate glomerular filtration rate from se- The data analysis and interpretation have 603 rum creatinine: a new prediction equation. Ann been done without the participation of these 13. Then C, Kowall B, Lechner A, et al. Plasma Intern Med 1999;130:461–470 organizations. copeptin is associated with type 2 diabetes in 27. Bjornstad P, Maahs DM, Jensen T, et al. Duality of Interest. No potential conflicts of men but not in women in the population-based Elevated copeptin is associated with atheroscle- interest relevant to this article were reported. KORA F4 study. Acta Diabetol 2015;52:103–112 rosis and diabetic kidney disease in adults with Author Contributions. G.V. and R.R. re- 14. Roussel R, El Boustany R, Bouby N, et al. type 1 diabetes. J Diabetes Complications 2016; searched data and wrote the manuscript. Plasma copeptin, AVP gene variants, and inci- 30:1093–1096 R.E.B., G.L., and K.M. researched data, contrib- dence of type 2 diabetes in a cohort from the 28. Meijer E, Bakker SJ, van der Jagt EJ, et al. uted to discussion, and reviewed and edited the community. J Clin Endocrinol Metab 2016;101: Copeptin, a surrogate marker of vasopressin, is manuscript. F.F., L.P., L.B., N.B., S.H., and M.M. 2432–2439 associated with disease severity in autosomal contributed to discussion and reviewed and 15. Ponte B, Pruijm M, Ackermann D, et al. Co- dominant polycystic kidney disease. Clin J Am edited the manuscript. G.V. is the guarantor peptin is associated with kidney length, renal Soc Nephrol 2011;6:361–368 of this work and, as such, had full access to all function, and prevalence of simple cysts in a 29. Boertien WE, Meijer E, Li J, et al.; Consor- the data in the study and takes responsibility population-based study. J Am Soc Nephrol tium for Radiologic Imaging Studies of Poly- for the integrity of the data and the accuracy 2015;26:1415–1425 cystic Kidney Disease CRISP. Relationship of of the data analysis. 16. Roussel R, Matallah N, Bouby N, et al. copeptin, a surrogate marker for arginine vaso- Prior Presentation. Parts of this study were Plasma copeptin and decline in renal function pressin, with change in total kidney volume and presented in abstract form at the 76th Scientific in a cohort from the community: the prospec- GFR decline in autosomal dominant polycystic Sessions of the American Diabetes Association, tive D.E.S.I.R. study. Am J Nephrol 2015;42:107– kidney disease: results from the CRISP cohort. New Orleans, LA, 10–14 June 2016. 114 Am J Kidney Dis 2013;61:420–429 17. Pikkemaat M, Melander O, Bengtsson 30. Torres VE, Chapman AB, Devuyst O, et al.; References Bostrom¨ K. Association between copeptin and TEMPO 3:4 Trial Investigators. Tolvaptan in 1. Bankir L, Bardoux P, Ahloulay M. Vasopressin declining glomerular filtration rate in people patients with autosomal dominant polycystic and diabetes mellitus. Nephron 2001;87:8–18 with newly diagnosed diabetes. The Skaraborg kidney disease. N Engl J Med 2012;367:2407– 2. Zerbe RL, Vinicor F, Robertson GL. Regulation Diabetes Register. J Diabetes Complications 2418 of plasma vasopressin in insulin-dependent di- 2015;29:1062–1065 31. Meijer E, Bakker SJ, de Jong PE, et al. abetes mellitus. Am J Physiol 1985;249:E317– 18. Velho G, Bouby N, Hadjadj S, et al. Plasma Copeptin, a surrogate marker of vasopressin, E325 copeptin and renal outcomes in patients with is associated with accelerated renal function de- 3. Taveau C, Chollet C, Waeckel L, et al. Vaso- type 2 diabetes and albuminuria. Diabetes cline in renal transplant recipients. Transplanta- pressin and hydration play a major role in the Care 2013;36:3639–3645 tion 2009;88:561–567 development of glucose intolerance and hepatic 19. Marre M, Jeunemaitre X, Gallois Y, et al. 32. Enhorning S, Hedblad B, Nilsson PM, steatosis in obese rats. Diabetologia 2015;58: Contribution of genetic polymorphism in the Engstrom G, Melander O: Copeptin is an inde- 1081–1090 renin-angiotensin system to the development pendent predictor of diabetic heart disease and 4. Bankir L, Bouby N, Ritz E. Vasopressin: a of renal complications in insulin-dependent di- death. Am Heart J 2015;169:549–556 novel target for the prevention and retardation abetes: Genetique de la Nephropathie Diabe- 33. Tasevska I, Enhorning¨ S, Persson M, Nilsson of kidney disease? Nat Rev Nephrol 2013;9:223– tique (GENEDIAB) study group. J Clin Invest PM, Melander O. Copeptin predicts coronary 239 1997;99:1585–1595 artery disease cardiovascular and total mortal- 5. Wang CJ, Grantham JJ, Wetmore JB. The me- 20. Hadjadj S, Pean´ F, Gallois Y, et al.; Genesis ity. Heart 2016;102:127–132 dicinal use of water in renal disease. Kidney Int France-Belgium Study. Different patterns of in- 34. Fenske W, Wanner C, Allolio B, et al.; Ger- 2013;84:45–53 sulin resistance in relatives of type 1 diabetic man Diabetes, Dialysis Study Investigators. care.diabetesjournals.org Velho and Associates 2295

Copeptin levels associate with cardiovascular induced insulin and glucagon release from the failure: study in Brattleboro rats. Life Sci 1999; events in patients with ESRD and type 2 diabe- perfused rat pancreas. Metabolism 2002;51: 65:991–1004 tes mellitus. J Am Soc Nephrol 2011;22:782–790 1184–1190 39. Groop PH, Thomas MC, Moran JL, et al.; 35. Koshimizu TA, Nakamura K, Egashira N, 37. Bouby N, Bachmann S, Bichet D, Bankir L. FinnDiane Study Group. The presence and se- Hiroyama M, Nonoguchi H, Tanoue A. Vasopres- Effect of water intake on the progression verity of chronic kidney disease predicts all- sin V1a and V1b receptors: from molecules to of chronic renal failure in the 5/6 nephrec- cause mortality in type 1 diabetes. Diabetes physiological systems. Physiol Rev 2012;92: tomized rat. Am J Physiol 1990;258:F973– 2009;58:1651–1658 1813–1864 F979 40. Lind M, Svensson AM, Kosiborod M, et al. 36. Abu-Basha EA, Yibchok-Anun S, Hsu WH. 38. Bouby N, Hassler C, Bankir L. Contribution Glycemic control and excess mortality in type 1 Glucose dependency of arginine vasopressin- of vasopressin to progression of chronic renal diabetes. N Engl J Med 2014;371:1972–1982