C-Terminal Proarginine Vasopressin is Associated with Disease Outcome and Mortality, but not with Delayed Cerebral Ischemia in Critically Ill Patients with an Aneurysmal Subarachnoid Hemorrhage: A Prospective Cohort Study

In a single-center, prospective, observational cohort study, we enrolled patients with aSAH admitted to the ICU of the Elisabeth Tweesteden hospital (Tilburg, the Netherlands) within 24 h after bleeding from November 2013 until April 2015. The study protocol was approved by the Medisch Ethische Toetsingscommissie Brabant (Tilburg, the Netherlands; trial number NL45096.008.13). Informed consent was obtained from participating patients. Inclusion criteria were adults ≥ 18 years of age, admittance to the ICU with an aSAH within 24 h after bleeding and a CT-proAVP index test on the day of ICU admission. Exclusion criteria for trial participation were recent (< 30 days) ischemic or hemorrhagic stroke, intracerebral hemorrhage without subarachnoid blood, head trauma, acute myocardial infarction, acute exacerbation of chronic obstructive pulmonary disease, sepsis or septic shock, acute pancreatitis, chronic heart failure, and liver cirrhosis. Diagnosis of aSAH was based on clinical symptoms (acute headache, focal neurological deficits, loss of consciousness), presence of blood on computerized tomography (CT) cerebrum or presence of xanthochromia in cerebral spinal fluid in combination with an aneurysm, confirmed by CT angiography or digital subtraction angiography (DSA) [1]. Diagnosis of DCI was based on acute clinical deterioration in the patient’s neurologic condition between three and 14 days after aSAH, assessed by a decrease of at least two points on the Glasgow Coma Scale sum score and/or by the development of new focal neurological deficits, and exclusion of other causes for neurological deterioration [9‐12]. In cases of suspected DCI, a CT brain perfusion, CT angiography, or DSA was performed. Other causes for neurological deterioration included increasing hydrocephalus, rebleeding of an aneurysm, epileptic seizure, severe infectious disease with associated decrease in level of consciousness, hypoglycemia (defined as serum glucose < 3 mmol/L), hyponatremia (defined as serum sodium < 125 mmol/L), and metabolic encephalopathy due to renal failure, as indicated by rapidly rising serum urea. We followed the Strengthening the Reporting of Observational Studies in Epidemiology Statement guidelines for reporting observational studies [26]. Included patients and excluded patients are described in the patient flow diagram. A control group consisted of 30 healthy volunteers, all hospital staff, with no vascular risk factors. The primary aim was the prediction of poor functional outcome after 1 year, and secondary aims were the prediction of 30-day and 1-year mortality and prediction of DCI by baseline CT-proAVP. Patients were contacted after 1 year by the research nurse (PV) for a questionnaire by telephone. This questionnaire included the Glasgow Outcome Scale (GOS) [27]. The GOS rated from death (one point) to symptom-free full recovery (five points). GOS scores were dichotomized in good and poor functional outcomes (GOS 4–5 vs. GOS 1–3, respectively). The research nurse was blinded for CT-proAVP levels.

Venous blood was drawn to measure CT-proAVP in the control group at the start of the study. Clinical data and laboratory results were collected on the first day of ICU admission in patients enrolled in the study. The initial CT-cerebrum was classified according to the modified Fisher scale [13, 15]. Blood samples were collected into clot-tubes at admittance. Serum was separated by centrifugation and stored in aliquots at − 80 °C, and Serum CT-proAVP levels were measured afterward using an automated immunofluorescent sandwich assay on a B.R.A.H.M.S. Kryptor Compact Plus analyzer (Thermo Fisher Scientific, Henningsdorf, Germany). The Kryptor measures the signal that is emitted from an immunocomplex by time-resolved amplified cryptate emission. CT-proAVP assays have a lower limit of detection of 0.69 pmol/L. The functional sensitivity (lowest value with an interassay coefficient of variation < 20% as described by the manufacturer) of 1.08 pmol/L. Imprecision of the assay was verified according to the Clinical & Laboratory Standards Institute Evaluation Protocol 17-A, using a low and high sample, measured for 5 days in triplicate. Intracoefficients of variation and intercoefficient of variation values were all ≤ 10% for CT-proAVP.

To study our hypothesis that CT-proAVP was useful as predictor for poor functional outcome at 1 year, a sample size of 93 patients will have 80% power to calculate sensitivity and specificity for CT-proAVP in patients with aSAH, using a χ² test with a 0.05 two-sided significance level. This power calculation was based on CT-proAVP levels in patients with aSAH in a prior study [21]. Normally distributed data were expressed as means (standard deviations); all nonnormally distributed data (Kolmogorov–Smirnov test p < 0.05) were expressed as medians (with interquartile ranges) or as number of patients (percentage), when appropriate. Patient characteristics and outcomes were compared using Mann–Whitney U-test for skewed distributed continuous variables and χ² test for categorical variables. The association between CT-proAVP or severity scores (WFNS and APACHE IV) and poor functional outcome at 1 year, mortality after 30 days and 1 year, and DCI was assessed by using area under the receiver operating characteristics (ROC) curves. Youden’s index analysis was applied to calculate optimal cutoff points. Youden’s index was represented by the following formula: J = sensitivity + specificity − 1. Sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios (LR+, LR−) were calculated for CT-proAVP and severity scores. CT-proAVP, WFNS, and APACHE IV were transformed to dichotomous variables (below or equal to and above the cutoff point) and included in univariable logistic regression models to study the effects on outcome, mortality and DCI. Sex, confirmed predictors of outcome and mortality (age, rebleeding), and DCI (modified Fisher scale) were also tested in univariable regression analysis. Variables that yielded p values < 0.10 were subsequently included in the multivariable regression analysis. Considering the low number of outcome measures in our study and to avoid overfitting of the model, CT-proAVP was tested with only a limited number of other variables. The model was checked for intercorrelations among the predictor variables by collinearity statistics. CT-proAVP levels were also log transformed to calculate the risk of poor functional outcome at 1 year in a logistic regression analysis formula. All tests were two-sided, and a p-value < 0.05 was considered statistically significant. All data were analyzed using a statistical software package (version 24; SPSS Inc., Chicago, IL).

During the recruitment period, 155 potentially eligible patients were admitted to the ICU with a presumed diagnosis of aSAH. CT-proAVP levels were measured the first day of admission and data of functional status after 1 year were obtained in 100 patients with SAH with a confirmed aneurysm. The patient flow diagram shows the flow of patients along with the primary end point of 1-year functional outcome (Fig. 1). Table 1 summarizes the clinical and laboratory data of these patients.

After 1 year, 45 patients had poor functional outcome and 55 had good functional outcome. Patients with poor functional outcome at 1 year had significantly higher CT-proAVP levels compared with patients with good functional outcome at 1 year (53.1 pmol/L [27.4–123.7] vs. 14.3 pmol/L [7.3–26.8], p < 0.001). ROC curves revealed high accuracy for CT-proAVP to identify patients with poor functional outcome at 1 year, area under the curve (AUC) 0.84 and 95% confidence interval (CI) 0.77–0.92, p < 0.001 (Table 2; Supplemental Fig. 2). When CT-proAVP was combined with WFNS or APACHE IV, the combination of APACHE IV and CT-proAVP yielded the highest AUC (Table 2). CT-proAVP and APACHE IV yielded the highest LR+ (Table 2). Eighty-two percent of the patients with both APACHE IV and CT-proAVP ≥ cutoff point had poor outcome at 1 year, and 90% of the patients with both APACHE IV and CT-proAVP < cutoff point had good outcome after 1 year (Supplemental Table 1). Univariable logistic regression analysis demonstrated that CT-proAVP ≥ 24.9 pmol/L and APACHE IV ≥ 44 points had the strongest association with increased risk of poor functional outcome at 1 year compared with patients with values below the cutoff point (Table 3). CT-proAVP, WFNS, APACHE IV and age were included in a multivariable logistic regression model. CT-proAVP ≥ 24.9 pmol/L proved to be a significant predictor for poor functional outcome at 1 year (odds ratio [OR] 8.04, 95% CI 2.97–21.75, p < 0.001) (Table 3). There was moderate correlation among predictor variables, variance inflation factors (VIF) of APACHE IV and WNFS were 4.74 and 4.04, respectively (Supplemental Table 2). The model was tested for interaction between APACHE IV and WFNS scores in a post-hoc analysis. An APACHE IV × WFNS interaction term made no significant contribution to the multivariable model (p = 0.097). The risk of poor functional outcome at 1 year could be calculated by the following logistic regression analysis formula: ln (p/1 − p) =  − 7.618 + 1.386α₁ + 0.047α₂, in which α₁ is ln (CT-proAVP level) and α₂ is APACHE IV score, p is the probability of poor functional outcome at 1 year, and p/1 − p is the odds of developing poor functional outcome after 1 year. WFNS score and age were not significant variables in the multivariable regression model.

Nineteen patients died of aSAH in 30 days and 25 patients died within 1 year. Nonsurvivors at 30 days and 1 year had significant higher concentrations of CT-proAVP the first day of admission than survivors (87.8 [30.1–228.8] vs. 18.4 pmol/L [9.7–39.9], p < 0.001 for 30-day mortality and 58.4 [29.0–163.8] vs. 18.4 [8.8–38.7], p < 0.001 for 1-year mortality). ROC curves revealed high accuracy for CT-proAVP to identify both patients with 30-day mortality (AUC 0.84, 95% CI 0.76–0.93, p < 0.001) and 1-year mortality (AUC 0.79, 95% CI 0.69–0.89, p < 0.001) (Table 4; (Supplemental Figs. 3, 4). The predictive value of CT-proAVP was lower than those of WFNS (30-day mortality) and APACHE IV (30-day and 1-year mortality). When CT-proAVP was combined with WFNS or APACHE IV, the combination of APACHE IV and CT-proAVP yielded the highest AUC (Table 4). APACHE IV yielded the highest LR+ for the prediction of 30-day and 1-year mortality (Table 4). All (100%) patients with both APACHE IV and CT-proAVP < cutoff point survived 30 days and 1 year (Supplemental Tables 3, 4). A smaller part (65% and 61%, respectively) of the patients with both APACHE IV and CT-proAVP ≥ cutoff point died in 30 days and 1 year (Supplemental Tables 3, 4). Univariable logistic regression analysis demonstrated that APACHE IV above the cutoff points had the strongest association with increased risk of 30-day and 1-year mortality compared with patients with values below the cutoff point (Table 5). CT-proAVP, WFNS, APACHE IV, age, and rebleeding were included in a multivariable logistic regression model for 30-day mortality and all, but rebleeding were included in further multivariable analysis for 1-year mortality. CT-proAVP ≥ 29.1 pmol/L and 27.7 pmol/L proved to be a significant predictor for 30-day and 1-year mortality (OR 9.31, 95% CI 1.55–56.07, p 0.015 and OR 5.15, 95% CI 1.48–17.93, p = 0.010), but not as strong predictor as APACHE IV in predicting 30-day and 1-year mortality (OR 18.27, 95% CI 1.19–281.53, p = 0.037 and OR 10.25, 95% CI 1.45–72.48, p = 0.020) (Table 5). There was a moderate correlation between predictor variables, APACHE IV and WFNS had the highest VIFs (Supplemental Tables 5, 6). The models were tested for interaction between APACHE IV and WFNS scores in a post-hoc analysis. An APACHE IV × WFNS interaction term made no significant contribution to the multivariable model of 30-day and 1-year mortality (p = 0.276 and p = 0.233, respectively).

Twenty-eight patients with aSAH experienced clinical signs of DCI during their ICU stay. Patients with DCI had significantly higher concentrations of CT-proAVP concentrations the first day of admission than patients without DCI (51 pmol/L [15.9–116.1] vs. 20.8 pmol/L [9.4–40.0], p 0.008). However, CT-proAVP had a low accuracy rate for identifying patients with DCI in the ROC analysis (AUC 0.67, 95% CI 0.55–0.79, p = 0.008) (Table 6; Supplemental Fig. 5). WFNS, APACHE IV, and modified Fisher scale had comparably low accuracy rates for predicting DCI, with low AUCs in ROC analysis (Table 6; Supplemental Fig. 5). An optimal cutoff point was calculated for CT-proAVP and modified Fisher scale. No optimal cutoff points could be calculated for WFNS and APACHE IV scores, and they were not tested in the multivariable model. CT-proAVP yielded a low LR+ for the prediction of DCI. Both CT-proAVP and modified Fisher scale were tested in a multivariable logistic regression model. CT-proAVP ≥ 29.5 pmol/L was not a significant predictor for DCI in a multivariable model adjusted for the modified Fisher scale (OR 2.51, 95% CI 0.96–6.56, p = 0.061) (Table 7). There was no correlation between CT-proAVP and modified Fisher scale (VIF 1.09) (Supplemental Table 7).