Prognostic and diagnostic significance of copeptin in acute exacerbation of chronic obstructive pulmonary disease and acute heart failure: data from the ACE 2 study

The Akershus Cardiac Examination (ACE) 2 Study was designed to assess the diagnostic and prognostic value of circulating biomarkers in patients admitted with acute dyspnea to Akershus University Hospital, Lørenskog, Norway. The primary aim of the ACE 2 study was to analyze the prognostic properties of secretoneurin, and a minimum sample size of 350 patients was originally calculated by power analysis for this purpose [21]. The method of patient recruitment and data collection has also been described in detail previously [22‐24]. Patients over the age of 18 years were eligible for inclusion during the first 24 h of admission if acute dyspnea was the primary cause for hospitalization as evaluated by the emergency department physician. Exclusion criteria were dementia and other causes precluding informed patient consent, disseminated malignant disease, acute myocardial infarction or coronary intervention, major surgery within the last 2 weeks, incomplete study blood sampling, and hemoglobin <10 g/dL. Consecutive patients were enrolled between 8 am and 2 pm Monday to Thursday. Clinical information was obtained from physicians on call, hospital records, and directly from the patients by dedicated study personnel who used standardized questionnaires. Echocardiography and spirometry results were registered from hospital records. The ACE 2 study was approved by the Norwegian Regional Committees for Medical and Health Research Ethics (REC) South East (#5.2008.2832) and conducted in agreement with the Declaration of Helsinki. All participants provided written informed consent prior to study enrolment.

The final diagnosis of the index hospitalization was established by two senior physicians working independently of each other, and discordant diagnoses were resolved by consensus. The two members of the adjudication committee had no knowledge of study biomarker levels, but they had access to all medical records, including follow-up data and cardiac biomarker measurements such as NT-proBNP and troponin T that were ordered by the treating physician. Patients were first classified into acute HF and non-HF related dyspnea, and then patients in the non-HF group were evaluated with respect to the AECOPD diagnosis. The acute HF diagnosis was determined by the European Society of Cardiology criteria [25], and the AECOPD diagnosis was based on the criteria defined by the Global initiative for Chronic Obstructive Lung Disease (GOLD) [26]. Discordant diagnoses were resolved by consensus. Survival status was recorded from electronic hospital records, which are synchronized with Statistics Norway, until the end of follow-up November 1st, 2012.

Standard biochemical work-up and arterial blood gas measurements were collected at admission. Glomerular filtration rate (GFR) was estimated by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula. Study blood sampling was performed by venipuncture and uniformly processed throughout the study period. Copeptin, N-terminal pro-B-type natriuretic peptide (NT-proBNP) and high-sensitivity cardiac troponin T (hs-TnT) were measured in samples obtained <24 h after hospital admission by commercially available assays: B-R-A-H-M-S Kryptor Copeptin assay by Thermo Fisher Scientific Inc., Clinical Diagnostics, BRAHMS GmbH, 16,761 Hennigsdorf, Germany; and proBNP II and troponin T hs STAT assays by Roche Diagnostics, Penzberg, Germany. Copeptin and hs-TnT were measured in serum while NT-proBNP was analyzed in plasma samples. The copeptin assay had a detection limit of 0.9 pmol/L, a functional sensitivity (inter-analysis variation <20%) above 2.0 pmol/L, and a normal reference range (2.5–97.5 percentile) of 0.9–14.9 pmol/L for healthy adults.

We report continuous variables as mean (± standard deviation [SD]) or median (quartile [Q] 1–3) depending on variable distribution. Differences between groups were compared by Student’s t test or Mann-Whitney U tests as appropriate. Binary data were compared by the chi-square test and are presented as absolute numbers and percentages. Positively skewed variables, including biomarkers, were log transformed by the natural logarithm to approach normal distribution and to reduce the effect of outliers in regression analysis. Variables associated with copeptin concentration were explored by Spearman’s rank correlation coefficient (rho) and linear regression analysis, and independent associations were determined by multivariate linear regression using stepwise forward selection of variables. Patient survival stratified by admission copeptin and NT-proBNP quartiles was analyzed by Kaplan-Meier plots and compared by the log-rank test. We identified factors associated with mortality by univariate Cox proportional hazard regression analysis. A basic multivariate Cox model of independent risk factors excluding biomarkers was constructed by stepwise forward selection based on the likelihood ratio criterion. The independent prognostic effect of each biomarker was determined by adjusting for the variables in the basic clinical risk model. The area under receiver operating curves (ROC-AUC) was used to ascertain the diagnostic and prognostic accuracies of biomarkers, while the value of adding biomarkers to the basic clinical risk models was investigated by calculating the category free net reclassification index (NRI). ROC-AUCs are presented with 95% confidence interval (CI) computed by bootstrap using 5000 iterations. We considered p < 0.05 to be statistically significant and statistical analyses were performed using SPSS for Windows version 22.0 (SPSS, Armonk, NY), STATA version 14 (Stata Corp LP, TX, USA), and R 3.3.3 (R Foundation for Statistical Computing, Vienna, Austria). NRI was calculated using the R package PredictABEL.

The prognostic properties of copeptin were analyzed in AECOPD and acute HF separately. After a median follow-up of 2.2 years (813 [356–996] days), 46% (66/143) of HF patients and 42% (35/84) of AECOPD patients had died. According to Kaplan-Meier estimates (Fig. 2) the risk of mortality during follow-up increased among acute HF patients if copeptin or NT-proBNP levels were elevated on hospital admission (p < 0.0001 by the log-rank test for both biomarkers). In contrast, only copeptin levels were associated with mortality among patients diagnosed with AEOCPD (Fig. 2; p < 0.0001 by the log-rank test). After adjustment for basic clinical risk factors, as identified by univariate screening (Additional file 1: Table S2), the risk of dying increased by 72% in AECOPD (HR 1.72 [1.21–2.45], p = 0.003) and 61% in acute HF (1.61 [1.25–2.09], p < 0.001) per log (ln) unit increment of copeptin by multivariate Cox analysis (Table 2). In comparison, one log (ln) unit increase of NT-proBNP increased the risk of mortality by 62% in acute HF (1.62 [1.27–2.06], p < 0.001), while no significant predictive effect was found in AECOPD (1.12 [0.88–1.42], p = 0.373). Neither copeptin nor NT-proBNP were independently associated with mortality in patients with dyspnea that was not related to CODP or HF (Additional file 1: Table S3). When copeptin and NT-proBNP was included in the same model, the predictive effect of copeptin was significant in patients with AECOPD (HR 1.79 [1.20–2.66], p = 0.004), but not in patients with acute HF (1.30 [0.96–1.76], p = 0.091). When copeptin was added to the basic clinical risk model, the category free net reclassification index (NRI) was positive in AECOPD (NRI 0.60 [0.19–1.02], p = 0.004) and acute HF (0.39 [0.06–0.71], p = 0.020). In the AECOPD group, the predicted risk of mortality decreased in 67% of survivors and increased in 63% of non-survivors with the model that included copeptin (Fig. 3). By ROC-AUC analysis, we could not find any statistical difference between the prognostic accuracy of copeptin and NT-proBNP in AECOPD (ROC-AUC 0.67 [0.55–0.79] vs. 0.56 [0.44–0.69], p = 0.111) or acute HF (0.66 [0.57–0.75] vs. 0.67 [0.58–0.76], p = 0.695). Adding NT-proBNP to the Cox regression model that already included copeptin or vice versa did not significantly alter the overall predictive accuracy of the model, as determined by ROC-AUC or category free NRI, in acute HF.

At admission, a large portion of AECOPD patients (35% [29/84]) and acute HF patients (64% [91/143] had copeptin concentrations that exceeded the upper reference limit (14.9 pmol/L) reported for healthy subjects by the manufacturer of the current assay. The median copeptin concentration was significantly higher among acute HF patients compared to patients with AECOPD and other causes of dyspnea (22.2 [10.2–47.9] vs. 8.8 [5.2–19.7] and 8.3 [4.3–18.2] pmol/L), but NT-proBNP discriminated acute HF from non-HF related dyspnea more accurately than copeptin (AUC 0.85 [95% CI 0.81–0.89] vs. 0.71 [0.66–0.77], p < 0.0001). We did not find any significant difference in the concentration of copeptin or NT-proBNP between patients with AECOPD and patients with other causes of dyspnea not related to HF. By multivariate linear regression analysis across all groups (Additional file 1: Table S1) increased copeptin concentrations were independently associated with increased hs-TnT, NT-proBNP, Na⁺, male gender and reduced eGFR (r² = 0.51). The correlation between copeptin and individual independent covariates were moderate for NT-proBNP (rho 0.60), hs-TNT (0.55), and eGFR (−0.52); and weak for male gender (0.27) and Na⁺ (0.24). While low Na⁺ levels were associated with lower levels of copeptin (Additional file 1: Figure S1), copeptin concentrations were measurable also among patients with hyponatremia (Na⁺ concentrations <137 mmol/L) in AECOPD (7.6 [2.7–16.0]) and acute HF (18.2 [6.3–52.6] pmol/L).