Diagnostic Accuracy of Copeptin in the Differential Diagnosis Of

ORIGINAL ARTICLE

Diagnostic Accuracy of Copeptin in the Differential Diagnosis of the Polyuria-polydipsia Syndrome: Downloaded from https://academic.oup.com/jcem/article-abstract/100/6/2268/2829618 by University of Alabama at Birmingham user on 06 September 2018 A Prospective Multicenter Study Katharina Timper, Wiebke Fenske, Felix Kühn, Nica Frech, Birsen Arici, Jonas Rutishauser, Peter Kopp, Bruno Allolio, Christoph Stettler, Beat Müller, Mira Katan, and Mirjam Christ-Crain

Clinic of Endocrinology, Diabetes and Metabolism, Department of Clinical Research (K.T., N.F., M.C.-C.), University Hospital Basel, CH-4031 Basel, Switzerland; Integrated Research and Treatment Center for Adiposity Diseases (W.F.), Leipzig University Medical Center, 04103 Leipzig, Germany; Division of Endocrinology, Diabetes and Clinical Nutrition (F.K., C.S.), University Hospital Bern–Inselspital, CH-3010 Bern, Switzerland; Department of Internal Medicine (B.Ar.), Spital Rheinfelden, CH-4310 Rheinfelden, Switzerland; University Clinic of Internal Medicine (J.R.), Kantonsspital Baselland, CH- 4101 Binningen, Switzerland; Division of Endocrinology, Metabolism and Molecular Medicine and Center for Genetic Medicine (P.K.), Northwestern University, Chicago, Illinois 60611; Department of Internal Medicine I, Endocrine and Diabetes Unit (B.Al.), University Hospital Würzburg, 97080 Würzburg, Germany; Division of Endocrinology, Diabetology and Metabolism, Medical University Clinic (B.M.), Kantonsspital Aarau, CH-5001 Aarau, Switzerland; and Department of Neurology (M.K.), University Hospital Zurich, CH-8091 Zurich, Switzerland

Context: The polyuria-polydipsia syndrome comprises primary polydipsia (PP) and central and nephrogenic diabetes insipidus (DI). Correctly discriminating these entities is mandatory, given that inadequate treatment causes serious complications. The diagnostic “gold standard” is the water deprivation test with assessment of arginine vasopressin (AVP) activity. However, test interpretation and AVP measurement are challenging.

Objective: The objective was to evaluate the accuracy of copeptin, a stable peptide stoichiometrically cosecreted with AVP, in the differential diagnosis of polyuria-polydipsia syndrome.

Design, Setting, and Patients: This was a prospective multicenter observational cohort study from four Swiss or German tertiary referral centers of adults Ͼ18 years old with the history of polyuria and polydipsia.

Measurements: A standardized combined water deprivation/3% saline infusion test was performed and terminated when serum sodium exceeded 147 mmol/L. Circulating copeptin and AVP levels were measured regularly throughout the test. Final diagnosis was based on the water deprivation/saline infusion test results, clinical information, and the treatment response.

Results: Fifty-five patients were enrolled (11 with complete central DI, 16 with partial central DI, 18 with PP, and 10 with nephrogenic DI). Without prior thirsting, a single baseline copeptin level Ͼ21.4 pmol/L differentiated nephrogenic DI from other etiologies with a 100% sensitivity and specificity, rendering a water deprivation testing unnecessary in such cases. A stimulated copeptin Ͼ4.9 pmol/L (at sodium levels Ͼ147 mmol/L) differentiated between patients with PP and patients with partial central DI with a 94.0% specificity and a 94.4% sensitivity. A stimulated AVP Ͼ1.8 pg/mL differentiated between the same categories with a 93.0% specificity and a 83.0% sensitivity.

Limitation: This study was limited by incorporation bias from including AVP levels as a diagnostic criterion.

Conclusion: Copeptin is a promising new tool in the differential diagnosis of the polyuria-polydipsia syndrome, and a valid surrogate marker for AVP. Primary Funding Sources: Swiss National Science Foundation, University of Basel. (J Clin Endocrinol Metab 100: 2268–2274, 2015)

ISSN Print 0021-972X ISSN Online 1945-7197 Abbreviations: AUC, under the curve; AVP, arginine vasopressin; CI, confidence interval; Printed in USA DI, diabetes insipidus; PP, primary polydipsia. Copyright © 2015 by the Endocrine Society Received December 23, 2014. Accepted March 10, 2015. First Published Online April 13, 2015

For related articles see pages 2275

2268 press.endocrine.org/journal/jcem J Clin Endocrinol Metab, June 2015, 100(6):2268–2274 doi: 10.1210/jc.2014-4507 doi: 10.1210/jc.2014-4507 press.endocrine.org/journal/jcem 2269

he polyuria-polydipsia syndrome comprises three ma- formed consent was obtained from all patients. The study was T jor entities: central (complete or partial) diabetes in- preregistered on ClinicalTrials.gov (NCT00757276). sipidus (DI), nephrogenic (complete or partial) DI, and Combined water deprivation/saline infusion test primary polydipsia (PP). Differentiating these entities is

All patients underwent a standardized water deprivation test Downloaded from https://academic.oup.com/jcem/article-abstract/100/6/2268/2829618 by University of Alabama at Birmingham user on 06 September 2018 important, given that inadequate treatment may lead to starting at 0800 h, without prior fluid restriction, according to serious complications, eg, profound hyponatremia. The the Robertson et al (4, 13) protocol, as long as baseline plasma diagnostic “gold standard” consists of a water deprivation sodium did not exceed 147 mmol/L. The test was stopped when test, after which urine osmolality should provide the cor- plasma sodium exceeded 147 mmol/L. At baseline and hourly during the test, blood pressure, pulse rate, and weight were mon- rect diagnosis. However, test interpretation is often chal- itored. Blood was sampled for measurement of plasma sodium, lenging, especially in distinguishing PP from partial forms osmolality, AVP, and copeptin, and urine was sampled for os- of DI, given that the kidney’s maximum concentrating molality determination. ability is often impaired due to a washout of the renal salt If plasma sodium levels increased greater than 147 mmol/L or gradient (1). Accordingly, in a recent study, the water de- were greater than 147 mmol/L at baseline, and urine osmolality remained less than 300 mmol/kg H O, the test was discontinued privation test led to a correct diagnosis in only 70% of 2 and a desmopressin challenge (2 ␮g i.v.; Ferring Pharmaceuticals, patients, including only 41% of patients with PP (2). Baar, Switzerland) was performed. Urine osmolality was measured Direct measurement of plasma vasopressin (AVP) be- before and 1 hour after desmopressin injection. If plasma sodium fore and after a thirsting period has been recommended for did not exceed 147 mmol/L by thirsting alone by 1300 h, patients better patient classification (3, 4). Although direct AVP received a 3% saline infusion at 0.1 mL/kg body weight/min and blood was sampled every 30 minutes thereafter for measurement of measurement led to a significantly better patient classifi- plasma sodium, osmolality, AVP, and copeptin. The test was ter- cation compared with urine osmolality measurement minated when plasma sodium exceeded 147 mmol/L. alone (4), this concept has not become accepted as the diagnostic standard due to several shortcomings. First, the Interpretation of the combined water deprivation/ normal range of plasma AVP in relation to plasma osmo- saline infusion test lality was originally established by studying a very small If upon thirsting alone, plasma sodium levels increased to greater group (5), whereas a recent larger study found a less close than 147 mmol/L and urine osmolality remained less than 300 mmol/kg H2O, complete DI was diagnosed. If urine osmolality in- association between the two parameters (2). Second, re- creased greater than 50% in response to desmopressin, complete liable plasma AVP measurement is cumbersome due to central DI was diagnosed, whereas a less than 50% increase diag- multiple preanalytical (6–10) and technical difficulties (6, nosed complete nephrogenic DI. If urine osmolality increased to 8, 10). In contrast, copeptin, the C-terminal glycoprotein greater than 300 mmol/kg H2O before plasma sodium exceeded 147 mmol/L, partial central or partial nephrogenic DI or PP was moiety of pro-AVP, is a stable surrogate marker of AVP diagnosed. In these cases, patients were further classified based on secretion. Copeptin recently has been suggested to im- AVP levels in relation to plasma osmolality levels (4, 5). If baseline prove the differential diagnosis of DI (2), to mirror water plasma AVP levels were greater than 2 pg/mL, whereas peak AVP deprivation and excess in healthy subjects (11), and to levels (at plasma sodium Ͼ147 mmol/L during fluid deprivation assess posterior pituitary function (12). and hypertonic saline infusion) were normal or high in relation to the concurrent plasma osmolality (4, 5), partial nephrogenic DI was The present study sought to prospectively investigate diagnosed. If baseline plasma AVP levels were less than 2 pg/mL and the performance of copeptin at baseline and after osmotic peak AVP levels were below normal in relation to the concurrent stimulation in the differential diagnosis of the entire spec- plasma osmolality, partial central DI was diagnosed. If baseline trum of the polyuria-polydipsia syndrome. plasma AVP levels were less than 2 pg/mL and peak AVP levels were normal in relation to the concurrent plasma osmolality, PP was diagnosed.

Materials and Methods Final diagnosis Because no established diagnostic “gold standard” exists, the Study cohort final diagnosis was made after careful, comprehensive evalua- In this prospective multicenter study, all adults referred to the tion by two independent experts in the field, blinded to copeptin Endocrine Units of the University Hospitals Basel, Aarau, and levels, according to the three diagnostic components outlined Bern, Switzerland, and Würzburg, Germany, from October below (2, 14). The final diagnosis was based on 1) the results of 2008 to August 2012 for diagnostic evaluation of the polyuria- the combined water deprivation/saline infusion test (including polydipsia syndrome were screened for eligibility. Inclusion cri- AVP levels) as described above, 2) additional features such as the teria were greater than 18 years of age and history of polyuria patient’s history including psychiatric disease, head trauma or (Ͼ40 mL/kg/d) and polydipsia. Exclusion criteria were osmotic surgery, family history, symptom onset, consistency of fluid in- diuresis due to diabetes mellitus or hypercalcemia, pregnancy, take, nocturia, and fluid preferences (14), as well as diagnostic uncorrected thyroid or adrenal insufficiency, and heart failure. findings (including anterior pituitary function tests, cranial mag- The local ethical committees approved the study. Written in- netic resonance imaging, evidence of autoimmune/inflammatory 2270 Timper et al Copeptin in Diabetes Insipidus J Clin Endocrinol Metab, June 2015, 100(6):2268–2274 dysfunction, or any neoplastic/infectious etiology of AVP defi- The same procedure was followed for stimulated and delta co- ciency) and 3) treatment response. If contradictory diagnoses peptin levels. Delta copeptin was calculated as stimulated co- resulted from these criteria, final diagnosis was based on treat- peptin at a plasma sodium level greater than 147 mmol/L Ϫ ment response. baseline copeptin, and delta AVP was calculated as stimulated Ϫ AVP at a plasma sodium level greater than 147 mmol/L base- Downloaded from https://academic.oup.com/jcem/article-abstract/100/6/2268/2829618 by University of Alabama at Birmingham user on 06 September 2018 Laboratory measurements line AVP. Routine measurements were performed immediately in non- Testing was two tailed, and P Ͻ .05 was considered statisti- frozen plasma and urine samples by automated biochemical cally significant. For statistical analysis, we used Stata 12.0 analyses in the University Hospital Central Laboratories. Blood (StataCorp LLP, College Station, TX). for copeptin analysis was sampled into EDTA tubes and centrifuged at 4°C, and plasma was stored at Ϫ20°C. Copeptin was measured in a single blinded batch by a chemiluminescence sand- wich immunoassay (BRAHMS CT-proAVP immunolumino- Results metric assay, Thermo Scientific Biomarkers, Hennigsdorf, Ger- many) developed by Morgenthaler et al (6) in 2006. The copeptin Baseline characteristics assay has a lower detection limit of 0.4 pmol/L, an intra-assay Fifty-five patients with the polyuria-polydipsia syn- variability of 3.7–2.5% at 2.14–514.4 pmol/L and an interassay drome were included in the study. Baseline characteristics variability of 22.9–3.6% at 0.58–902.9 pmol/L (6, 15). For measurement of copeptin only 50 ␮L of serum or plasma are needed, are shown in Table 1. Patients in the complete nephrogenic no preanalytical procedures are required, and the assay takes DI group were older, had a higher plasma osmolality, only 2–3 hours (6, 15). Blood for AVP measurement was sampled higher plasma sodium levels, and a higher lithium intake in chilled heparin tubes and immediately centrifuged at 4°C. (P ϭ .004, P ϭ .0008 and P ϭ .0057) whereas 24 hours Plasma was stored at Ϫ80°C until batch analysis. AVP levels fluid intake was higher in complete central DI patients were determined by RIA in Peter Kopp’s laboratory, University of Chicago, using the methodology of Robertson et al (4). (P ϭ .0037). Otherwise, there were no statistically significant differences in the baseline characteristics between the Statistics entities of the polyuria-polydipsia syndromes. Discrete variables are expressed as numbers (percentages), continuous variables as medians (interquartile range), and hor- Baseline AVP and copeptin levels mone concentration ranges as minimum–maximum. Frequency comparisons for categorical baseline measurements were per- Baseline AVP levels ranged from 3.0–16.7 pg/mL in formed by ␹2 test; to compare medians, we used the Whitney U nephrogenic DI patients (complete, 3.0–16.7 pg/mL; par- test or Kruskal-Wallis test (with Bonferroni correction for mul- tial, 4.2–12.1 pg/mL), from 0.5–2.4 pg/mL in central DI tiple comparison performed as needed). patients (complete, 0.5–1.4 pg/mL; partial, Ͻ0.5–2.4 pg/ For the association between AVP and copeptin levels, Spear- mL), and from 0.5–2.7 pg/mL in PP patients (Figure 1A). man correlation analyses were performed. To assess the discriminatory accuracy of baseline copeptin Baseline copeptin levels ranged from 21.4–117.0 pmol/L levels for the differential diagnosis of the polyuria-polydipsia in patients with nephrogenic DI (complete, 38.7–117.0 syndrome, we calculated the area under the receiver-operating- pmol/L; partial, 21.4–26.6 pmol/L), from 0.7–5.1 pmol/L characteristic curve for each of the five diagnoses, ie, complete or in patients with central DI (complete, 0.7–3.4 pmol/L; par- partial central DI, complete or partial nephrogenic DI, or PP. We then compared the area under the curve (AUC) of copeptin to tial, 0.9–5.1 pmol/L), and from 0.9–13.5 pmol/L in pa- that of AVP for each diagnosis, using the DeLong Method (16). tients with PP (Figure 1B).

Table 1. Baseline Patient Characteristics by Polyuria-Polydipsia Syndrome Entity (n ϭ 55)

Complete Partial Nephrogenic Complete Characteristic Central DI Partial Central DI Primary Polydipsia DI Nephrogenic DI

n 11161846 Age, y, median (IQR) 44.0 (38.0–58.0) 45.5 (29.0–55.8) 36.0 (26.3–45.8) 48.5 (44.0–57.3) 62.5 (57.3–64.8) Sex, F/M (% F) 5/6 (45%) 9/7 (56%) 13/5 (72%) 1/3 (25%) 4/2 (67%) BMI, kg/m2, median (IQR) 24.5 (22.2–29.7) 26.3 (21.7–31.6) 25.2 (19.7–27.4) 34.9 (29.1–39.9) 25.5 (24.5–26.4) Fluid intake, L/d, median (IQR) 10.0 (6.0–11.5) 6.0 (4.0–8.0) 5.0 (4.0–7.8) 5.5 (4.0–7.3) 4.8 (4.1–5.4) Nighttime fluid intake, yes/no (% yes) 10/1 (91%) 11/5 (69%) 7/11 (39%) 2/2 (50%) 3/3 (50%) Plasma sodium, mmol/L, median (IQR) 142.0 (141.5–146.0) 140.5 (138.0–143.0) 139.5 (138.3–141.0) 141.0 (140.0–141.5) 147.5 (144.8–149.5)

Plasma osmolality, mmol/kg H2O, median (IQR) 298.0 (291.0–303.5) 288.5 (280.8–292.3) 289.5 (286.0–299.0) 300.0 (298.5–301.0) 331.5 (315.8–342.0) Hypovolemic/euvolemic/hypervolemic status (% euvolemic) 3/8/0 (73%) 0/16/0 (100%) 2/16/0 (89) 0/3/0 (100%) 2/2/0 (50%) SBP, mm Hg, median (IQR) 127.0 (120.0–145.0) 127.5 (120.8–133.5) 116.5 (106.3–131.5) 121.5 (113.5–127.5) 138.5 (120.5–140.0) DBP, mm Hg, median (IQR) 88.0 (77.5–92.0) 76.5 (74.8–83.5) 74.0 (70.0–83.0) 81.0 (75.0–82.8) 85.0 (77.8–90.8) Medication: Present or past lithium treatment, yes/no (% yes) 0/11 (0) 1/15 (6%) 1/17 (6%) 2/2 (50%) 5/1 (83%)

Abbreviations: BMI, body-mass index; BW, body weight; DI, diabetes insipidus; DBP, diastolic blood pressure; F, female; IQR, interquartile range; M, male; SBP, systolic blood pressure. Discrete variables are expressed as numbers (percentages) and continuous variables as medians (IQR). doi: 10.1210/jc.2014-4507 press.endocrine.org/journal/jcem 2271

lation test was performed. Therefore, copeptin and AVP levels in these patients did not differ from baseline levels. In patients with partial nephro- genicDI,stimulatedAVPlevelsranged Downloaded from https://academic.oup.com/jcem/article-abstract/100/6/2268/2829618 by University of Alabama at Birmingham user on 06 September 2018 from 2.5–7.7 pg/mL, and stimulated copeptin levels ranged from 19.3–39.9 pmol/L.

Correlation of copeptin and AVP levels

Figure 1. Baseline AVP and copeptin plasma levels in the differential diagnosis of the polyuria- Baseline and stimulated copeptin polydipsia syndrome. Box plots depicting interquartile ranges with medians and whiskers correlated with corresponding AVP depicting minimal and maximal values for baseline AVP (A) and copeptin (B) levels without prior plasma levels (r ϭ 0.68, P Ͻ .001; r ϭ thirsting in patients with partial and complete central DI, patients with PP, and patients with 0.81, P Ͻ .001, respectively). Delta partial and complete nephrogenic DI. Cutoffs for best discrimination between nephrogenic vs nonnephrogenic DI diagnoses for AVP and copeptin are shown. The overall P value refers to copeptin (ie, stimulated copeptin– Kruskal-Wallis test results across all patient subgroups. For all two-subgroup comparisons, Mann- baseline copeptin) correlated with Whitney [ital]U test tests were used and the P value was Bonferroni adjusted. delta AVP (r ϭ 0.70, P Ͻ .001).

Osmotically stimulated AVP and copeptin during Baseline AVP and copeptin levels in differential the combined water deprivation/saline infusion diagnosis of the polyuria-polydipsia syndrome test Without prior thirsting, an AVP level of at least 3.0 Osmotically stimulated plasma AVP levels (at a sodium pg/mL or a copeptin level of at least 21.4 pmol/L at base- concentration of greater than 147 mmol/L during the com- line (Figure 1, A and B) differentiated all patients with bined water deprivation/saline infusion test) ranged from nephrogenic DI from patients with other diagnoses with 0.5–1.2 pg/mL and 0.8–1.8 pg/mL in patients with com- 100% sensitivity and specificity (respective AUCs [95% plete and partial central DI, and from 0.7–12.2 pg/mL in confidence intervals; CIs] 1.0, [1.0–1.0] and 1.0, [1.0– patients with PP (Figure 2A). Stimulated copeptin levels 1.0], comparison of AUCs, P ϭ not statistically significant ranged from 0.7–3.3 pmol/L and 0.7–7.0 pmol/L, respec- (n.s.). A baseline copeptin level of at least 2.9 pmol/L dif- tively, in patients with complete and partial central DI, and ferentiated between patients with central DI (complete from 3.7–24.4 pmol/L in patients with PP (Figure 2B). In and partial) and other diagnoses with 82% sensitivity and five of six patients with complete nephrogenic DI, plasma 78% specificity (AUC [95% CI], 0.81 [0.68–0.93]). In sodium levels were already greater than 147 mmol/L at base- comparison, a baseline AVP of at least 1.8 pg/mL differen- line and consequently, according to the protocol, no stimu- tiated between these diagnoses with 54% sensitivity and 89% specificity (AUC [95% CI], 0.69 [0.55–0.84]; comparison of AUCs, P ϭ n.s.).

Osmotically stimulated AVP and copeptin levels in differential diagnosis of the polyuria- polydipsia syndrome An osmotically stimulated AVP level of at least 1.8 pg/mL had 83% sensitivity and 93% specificity (AUC [95% CI], 0.89 [0.76–1.00]), Figure 2. Osmotically stimulated AVP and copeptin plasma levels in the differential diagnosis of whereas a stimulated copeptin level the polyuria-polydipsia syndrome. Box and whisker plots with medians and minimal and maximal of at least 4.9 pmol/L had 94% sen- values for stimulated AVP (A) and copeptin (B) plasma values at a plasma sodium level Ͼ147 mmol/L are depicted for patients with complete and partial central DI and for patients with PP. sitivity and specificity (AUC [95% Osmotic stimulation was provided by a combined water deprivation and saline infusion test. The CI], 0.98 [0.95–1.00]) to differenti- cutoffs for best discrimination between PP vs central DI for AVP and copeptin are depicted. The ate between PP and partial central DI overall P value refers to Kruskal-Wallis test results across all patient subgroups. For all two- ϭ subgroup comparisons, Mann-Whitney [ital]U tests were used and the P value was Bonferroni (comparison of AUCs, P n.s.). Re- adjusted. garding the discrimination of PP vs 2272 Timper et al Copeptin in Diabetes Insipidus J Clin Endocrinol Metab, June 2015, 100(6):2268–2274

Table 2. Plasma Copeptin and AVP for the Discrimination of Polyuria-Polydipsia Syndromes: AUCs and Proposed Diagnostic Cutoffs With Sensitivity and Specificity

Other Polyuria-Polydipsia Nephrogenic DI Versus Other Syndrome Diagnoses Primary Polydipsia Primary Polydipsia Downloaded from https://academic.oup.com/jcem/article-abstract/100/6/2268/2829618 by University of Alabama at Birmingham user on 06 September 2018 Polyuria-Polydipsia Syndrome Versus Complete Versus Partial Central Versus Central (Partial Variable Diagnoses, AUC (95% CI) Central DI, AUC (95% CI) DI, AUC (95% CI) and Complete) DI, AUC (95% CI)

Baseline plasma copeptin cutoffs, 1.00 (1.00–1.00) 0.74 (0.59–0.88) 0.68 (0.49–0.88) 0.70 (0.53–0.88) pmol/L (sensitivity/specificity, %) Cutoff: >21.4 (100%/100%) Cutoff: Ն0.9 (100%/9%) P ϭ n.s. Cutoff: Ն0.9 (100%/4%) Cutoff: >2.6 (64%/64%) Cutoff: >2.9 (72%/78%) Cutoff: Ն3.7 (43%/100%) Cutoff: Ն6.2 (17%/100%) Stimulated plasma copeptin cutoffs, — 0.90 (0.81–0.98) 0.98 (0.94–1.00) 0.99 (0.94–1.00) pmol/L (sensitivity/specificity, %) Cutoff: Ն0.8 (98%/9%) Cutoff: Ն3.7 (100%/81%) Cutoff: Ն3.7 (100%/89%) Cutoff: >2.6 (89%/82%) Cutoff: >4.9 (94%/94%) Cutoff: >4.9 (94%/96%) Cutoff: Ն3.7 (71%/100%) Cutoff: Ն7.2 (78%/100%) Cutoff: Ն7.2 (72%/100%) Baseline plasma AVP cutoffs, 1.00 (1.00–1.00) 0.66 (0.51–0.80) 0.51 (0.31–0.71) 0.53 (0.35–0.72) pg/mL (sensitivity/specificity, %) Cutoff: >3.0 (100%/100%) Cutoff: Ն0.6 (89%/9%) P ϭ n.s. P ϭ n.s. Cutoff: >1.2 (52%/82%) Cutoff: Ն1.6 (43%/100%) Stimulated plasma AVP cutoffs, — 0.89 (0.80–0.97) 0.89 (0.76–1.00) 0.90 (0.78–1.00) pg/mL (sensitivity/specificity, %) Cutoff: Ն0.8 (100%/18%) Cutoff: Ն0.9 (94%/13%) Cutoff: Ն0.7 (100%/7%) Cutoff: >1.2 (80%/82%) Cutoff: >1.8 (83%/93%) Cutoff: >1.8 (83%/96%) Cutoff: Ն1.3 (77%/100%) Cutoff: Ն1.9 (78%/100%) Cutoff: Ն1.9 (78%/100%)

Abbreviation: n.s. ϭ not statistically significant (P Ͼ .05). Potential diagnostic cutoffs for baseline and osmotically stimulated (ie, at a plasma sodium level Ͼ147 mmol/L) AVP are given with their respective sensitivity and specificity in parentheses. Proposed clinically useful cutoffs for copeptin and AVP for each differential diagnoses are shown in bold. central (complete and partial) DI, a stimulated AVP level Discussion of at least 1.8 pg/mL had 83% sensitivity and 96% specificity (AUC [95% CI], 0.91 [0.78–1.00]). A stimulated The results of this prospective international multicenter study show that plasma copeptin is a reliable marker to copeptin level of at least 4.9 pmol/L had 94% sensitivity discriminate between different entities of the polyuria- and 96% specificity (AUC [95% CI], 0.99 [0.97–1.00]; polydipsia syndrome and correlates well with AVP plasma comparison of AUCs, P ϭ n.s.) to differentiate PP from levels. The study had the following main findings. First, a central (partial and complete) DI. single baseline copeptin measurement without prior fluid Table 2 summarizes proposed baseline and stimulated restriction reliably discriminates patients with nephro- AVP and copeptin cutoffs in the differential diagnosis of genic DI from patients with all other entities of the poly- the polyuria-polydipsia syndrome. uria-polydipsia syndrome, rendering a water deprivation test unnecessary in the former subgroup. Second, a base- Diagnostic performance of the combined water line copeptin measurement allowed a good but not un- deprivation/saline infusion test equivocal identification of patients with complete central Compared with the final diagnosis, the combined water DI. Third, osmotically stimulated copeptin values differ- deprivation/saline infusion test including AVP measure- entiated between PP and partial central DI with 94% sen- ment allowed a definite diagnosis to be established in all 11 sitivity and specificity, which was even higher when pa- patients with complete central DI, in 15 of 16 patients with tients with complete central DI were included in the partial central DI, in eight of 18 patients with PP, and in analysis. Given that treatment does not differ in patients all 10 patients with nephrogenic DI. Altogether, 44 of 55 with partial vs complete central DI, a strict differentiation patients (80%) were diagnosed correctly. However, this between these two entities is generally of minor clinical approach led to a definite diagnosis in only 44% of pa- relevance. Overall, with the help of copeptin, a definite tients with PP. diagnosis was achieved in 96% of patients. In comparison, the combined water deprivation/saline Our results show that the performance of copeptin and infusion test including copeptin measurement led to a def- AVP are quite similar. The major advantage of using co- inite diagnosis in 10/11 patients with complete central DI, peptin instead of AVP in the diagnostic workup of patients in 15/16 patients with partial central DI, in 18/18 patients with the polyuria-polydipsia syndrome is, that copeptin is with PP, and in 10/10 patients with nephrogenic DI. more reliable and much easier to measure in clinical rou- Hence, a definite diagnosis could be established in 53/55 tine due to a standardized, fast assay that can be used all patients (96%) through copeptin measurement and the over the world with only a small interassay variability and combined water deprivation/saline infusion test. due to very little preanalytical challenges. To reduce po- doi: 10.1210/jc.2014-4507 press.endocrine.org/journal/jcem 2273 tential variability in AVP-measurement, due to the known ond, a single baseline copeptin measurement could not preanalytical problems as observed in other studies (2), in unequivocally diagnose complete central DI in our study, our study AVP samples have been obtained under stan- most probably due to differences in protocols, given that dardized optimized conditions and AVP was measured in baseline copeptin values in the Fenske study were obtained Chicago using the methodology of Robertson et al (4). after overnight thirsting, whereas in our study, baseline Downloaded from https://academic.oup.com/jcem/article-abstract/100/6/2268/2829618 by University of Alabama at Birmingham user on 06 September 2018 The indirect water deprivation test has been the diag- copeptin values were measured at 0800 h with prior ad nostic “gold standard” to differentiate between PP and libitum fluid intake. partial central DI. However, in a recent study, this ap- The strengths of our study include the evaluation of the proach provided a classification agreeing with the final diagnostic accuracy of copeptin in patients with all entities diagnosis in only 41% of patients (2). Moreover, in that of the polyuria-polydipsia syndrome, encompassing a rel- study, direct AVP measurement combined with the water evant number of patients with nephrogenic DI, and a test deprivation test correctly diagnosed only 40% of patients protocol ensuring a significant osmotic stimulus in all pa- with PP who attained a serum osmolality greater than 290 tients. However, our study had several limitations. First, mmol/kg H2O after 16 hours (2). In our study, direct AVP the final diagnosis, based on the combined water depri- measurement according to the protocol of Robertson et al vation and saline infusion test with AVP measurements, (4, 5) led to a definite diagnosis of PP in only 44% of clinical features, and treatment response, was not stan- patients with this entity, although high plasma sodium dardized. However, because no validated “gold standard” (Ͼ147 mmol/L) or plasma osmolality (Ͼ295 mmol/kg is available, every diagnostic study struggles with this lim-

H2O) levels, or both, were obtained in all patients. Nine of itation. Our diagnostic approach has been assessed pre- 18 patients with PP were falsely diagnosed with partial viously and also is commonly used in clinical routine (2). central DI and one of 18 patients with PP was incorrectly Second, by evaluating the diagnostic accuracy of AVP, we diagnosed with partial nephrogenic DI. The poor perfor- introduced an incorporation bias, given that AVP levels mance of AVP in differentiating PP may have different contributed to our final diagnosis. However, this factor reasons. The relationship defined for AVP and plasma may have led only to a bias toward the null hypothesis for osmolality that serves as a diagnostic tool in the combined the performance of copeptin (underestimation of diagnos- water deprivation/saline infusion test is based on a small tic accuracy) in comparison with AVP (overestimation of number (n ϭ 11) of individuals in the original publication diagnostic performance). Third, copeptin cutoffs were (5). Furthermore, and as noted previously (2), the original generated in the same cohort used to evaluate its diagnos- publications (5, 17, 18) did not describe how the AVP tic accuracy. Overall, the proposed copeptin cutoffs must reference range in relation to plasma sodium was estab- be validated prospectively in multicenter studies including lished. In addition, reliable AVP measurement is cumber- sufficiently high numbers of patients of all entities of the some due to preanalytical difficulties, given that the ma- polyuria-polydipsia syndrome. ture hormone is highly unstable (6), largely attached to In summary, measurement of baseline copeptin levels platelets (7), and rapidly cleared from plasma (6, 8–10). without prior fluid restriction allows unequivocal diag- By contrast, copeptin, which is cosecreted stoichiometri- nosis of patients with nephrogenic DI. Measurement of cally with AVP (11, 12, 19), is more stable, and can be circulating copeptin levels upon a sufficient osmotic stim- easily measured in serum or plasma (6, 15). ulus provides a new tool to discriminate between complete Our findings support and expand on earlier findings by and partial central DI vs PP. Copeptin therefore improves Fenske et al (2) concerning the diagnostic utility of copep- and simplifies differential diagnosis in the polyuria-poly- tin in the polyuria-polydipsia syndrome. However, some dipsia syndrome. of our results also differ. First, the diagnostic accuracy of stimulated plasma copeptin levels to differentiate between patients with PP vs partial central DI was higher in our Acknowledgments study. This observation might be explained by the fact that in the Fenske study, almost half of the patients had an We thank Ursula Schild, Vreni Wyss, and Ursula Duerring for technical support and performance of the water deprivation insufficient increase in serum osmolality at the end of the tests, and Mary Beth Gaskill for technical assistance with AVP water deprivation test and therefore had a limited stimulus measurements. The manuscript was edited by Robert Marlowe. for copeptin and AVP secretion. This observation high- This study was investigator-initiated and mainly supported by lights the importance of a sufficient increase in serum os- the participating centers’ research funds and the Swiss National molality and sodium levels, which in our study, was Foundation. Thermo Scientific Biomarkers, Hennigsdorf, Ger- achieved by administration of a hypertonic saline infusion many, manufacturer/developer of copeptin assays, performed subsequent to the water deprivation test if necessary. Sec- the study copeptin testing using reagents supplied gratis. No 2274 Timper et al Copeptin in Diabetes Insipidus J Clin Endocrinol Metab, June 2015, 100(6):2268–2274 funding source had any other involvement in the collection, anal- 4. Zerbe RL, Robertson GL. A comparison of plasma vasopressin mea- ysis, or interpretation of the data or the decision to approve surements with a standard indirect test in the differential diagnosis publication of the finished manuscript. The principal investiga- of polyuria. N Engl J Med. 1981;305(26):1539–1546. 5. Baylis PH, Robertson GL. Plasma vasopressin response to hyper- tors had complete and final control of the study design and con- tonic saline infusion to assess posterior pituitary function. J R Soc duct, database, statistical analysis, publication decisions, and Med. 1980;73(4):255–260. Downloaded from https://academic.oup.com/jcem/article-abstract/100/6/2268/2829618 by University of Alabama at Birmingham user on 06 September 2018 manuscript content. 6. Morgenthaler NG, Struck J, Alonso C, Bergmann A. Assay for the measurement of copeptin, a stable peptide derived from the precur- Address all correspondence and requests for reprints to: Mir- sor of vasopressin. Clin Chem. 2006;52(1):112–9. jam Christ-Crain, MD, PhD, Department of Clinical Research, 7. Preibisz JJ, Sealey JE, Laragh JH, Cody RJ, Weksler BB. Plasma and platelet vasopressin in essential hypertension and congestive heart University Hospital Basel, Clinic for Endocrinology, Diabetes failure. Hypertension. 1983;5(Pt 2):I129–I138. and Metabolism, Petersgraben 4, CH-4031 Basel, Switzerland. 8. Robertson GL, Mahr EA, Athar S, Sinha T. Development and clin- E-mail: [email protected]. ical application of a new method for the radioimmunoassay of ar- Author Contributions: K.T. and M.K. were involved in the ginine vasopressin in human plasma. J Clin Invest. 1973;52(9): study design, study coordination, recruitment of patients, and 2340–2352. statistical analysis, and they drafted the manuscript. W.F. re- 9. Baumann G, Dingman JF. Distribution, blood transport, and deg- cruited patients and drafted the manuscript. F.K., N.F., and radation of antidiuretic hormone in man. J Clin Invest. 1976;57(5): B.Ar. were involved in the study coordination and recruited pa- 1109–1116. tients. P.K., B.Al., C.S., J.R., and B.M. were involved in study 10. Czaczkes JW, Kleeman CR, Koenig M. Physiologic Studies of An- coordination, gave staff support, and drafted the manuscript. tidiuretic Hormone by Its Direct Measurement in Human Plasma. J Clin Invest. 1964;43:1625–1640. M.C.-C. designed and coordinated the study, drafted the man- 11. Szinnai G, Morgenthaler NG, Berneis K, et al. Changes in plasma uscript, and gave financial and staff support. All authors read copeptin, the c-terminal portion of arginine vasopressin during wa- and approved the final manuscript. Robert Marlowe edited the ter deprivation and excess in healthy subjects. J Clin Endocrinol manuscript. Metab. 2007;92(10):3973–3978. This study was registered in ClinicalTrials.gov as trial number 12. Katan M, Morgenthaler NG, Dixit KC, et al. Anterior and posterior NCT00757276. pituitary function testing with simultaneous insulin tolerance test This work was supported by grants from the University of and a novel copeptin assay. J Clin Endocrinol Metab. 2007;92(7): Basel (Nachwuchsförderung 2008) to K.T. and by a grant from 2640–2643. the Swiss National Science Foundation, No. PP00P3-12346, to 13. Milles JJ, Spruce B, Baylis PH. A comparison of diagnostic methods to differentiate diabetes insipidus from primary polyuria: A review M.C.-C. of 21 patients. Acta Endocrinol. 1983;104(4):410–416. Disclosure Summary: W.F., B.M., and M.C.-C. have received 14. Fenske W, Allolio B. Clinical review: Current state and future per- speaker honoraria from Thermo Scientific Biomarkers, the co- spectives in the diagnosis of diabetes insipidus: A clinical review. peptin assay developer/manufacturer. Thermo Scientific Bio- J Clin Endocrinol Metab. 2012;97(10):3426–3437. markers funded Robert Marlowe’s editing of the manuscript. 15. Morgenthaler NG, Struck J, Jochberger S, Dünser MW. Copeptin: The other authors have nothing to disclose. Clinical use of a new biomarker. Trends in endocrinology and metabolism: Trends Endocrinol Metab. 2008;19(2):43–49. 16. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic References curves: A nonparametric approach. Biometrics. 1988;44(3):837– 845. 1. Robertson GL. Diabetes insipidus. Endocrinol Metab Clin North 17. Zerbe R, Stropes L, Robertson G. Vasopressin function in the syn- Am. 1995;24(3):549–572. drome of inappropriate antidiuresis. Annu Rev Med. 1980;31:315– 2. Fenske W, Quinkler M, Lorenz D, et al. Copeptin in the differential 327. diagnosis of the polydipsia-polyuria syndrome—Revisiting the di- 18. Robertson GL, Aycinena P, Zerbe RL. Neurogenic disorders of os- rect and indirect water deprivation tests. J Clin Endocrinol Metab. moregulation. Am J Med. 1982;72(2):339–353. 2011;96(5):1506–1515. 19. Balanescu S, Kopp P, Gaskill MB, Morgenthaler NG, Schindler C, 3. Baylis PH, Gaskill MB, Robertson GL. Vasopressin secretion in Rutishauser J. Correlation of plasma copeptin and vasopressin con- primary polydipsia and cranial diabetes insipidus. Q J Med. 1981; centrations in hypo-, iso-, and hyperosmolar States. J Clin Endo- 50(199):345–358. crinol Metab. 2011;96(4):1046–1052.