If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address for reprints: Yanqin Cui, MD, Cardiac Intensive Care Unit, Heart Center, Guangzhou Women and Children's Medical Center, Guangzhou Medical University, No. 9, Rd Jinsui, Guangzhou, Guangdong, China.
Cardiac Intensive Care Unit, Heart Center, Guangzhou Women and Children's Medical Center, Guangzhou Medical University, Guangzhou, China
To assess the relationship between N-terminal pro–brain natriuretic peptide (NT-proBNP) levels at different time points and early outcome, and to evaluate the reliability of NT-proBNP level as a predictor of early outcome after surgery in a large series of children with congenital heart disease (CHD).
A retrospective observational study involving 363 consecutive children with CHD was used. Plasma NT-proBNP records were obtained for each patient before and 1, 12, and 36 hours after surgery. The specificity, sensitivity, and prediction value of NT-proBNP in predicting early postoperative outcomes were determined.
Analyses confirmed that time-varying NT-proBNP level, particularly 1-hour postoperative levels, had prognostic value on the prediction of prolonged duration of mechanical ventilation, intensive care unit (ICU) stay, and inotropic therapy. Joint modeling analyses of a linear mixed effects model for NT-proBNP from before to 36 hours after surgery and generalized linear models for the duration of the mechanical ventilation, ICU stay, and inotropic therapy showed that a 1% increase in NT-proBNP was associated with 5.5%, 3.9%, and 3.5% relative increases in expected duration of mechanical ventilation, ICU stay, and inotropic therapy, respectively; related P values were .001, .001, and .01, respectively.
After CHD surgery, the perioperative NT-proBNP levels might be powerful markers to identify subjects at higher risk for worse outcome.
We found that perioperative NT-proBNP levels, particularly 1-hour postoperative levels, were correlated with duration of mechanical ventilation, ICU stay, and inotropic therapy. We used joint models to quantify the degree of the correlation between perioperative NT-proBNP level and the length of mechanical ventilation, ICU stay, and inotropic therapy. It will help clinicians to predict early outcome and plan their therapeutic protocols more quickly after surgery.
Although great progress has been made, surgical intervention for the treatment of children with congenital heart disease (CHD) is still carrying significant early and late morbidity and mortality. Making the continued search for improved prognostic biomarkers besides commonly measured clinical and physiologic parameters is of increasing importance of children admitted to intensive care units (ICUs) after surgery for CHD. Accumulating evidence suggests that both active B-type natriuretic peptide (BNP) and inactive N-terminal pro-BNP (NT-proBNP) have been shown to have prognostic value in adults with cardiac diseases (eg, chronic congestive heart failure and acute myocardial infarction), and to predict morbidity and major adverse cardiac events after adult heart surgery.
Recently, more concerted efforts for the prognostic value of BNP/NT-proBNP after surgical intervention for CHD in pediatrics have also occurred.
found that BNP level of 1 or greater at 24 hours after cardiac surgery in neonates was correlated with poor postoperative outcome, including low cardiac output syndrome, fewer ventilator-free days, and 6-month major adverse cardiac events. Niedner et al
reported similar results in a prospective comparative study on 19 infants aged 1 to 12 months with CHD undergoing complete repair. In a prospective study on 587 children undergoing cardiac surgery, Cantinotti et al
there are much fewer data on the prognostic value of perioperative NT-proBNP levels in children after surgical repair of structural congenital heart defects, particularly large-scale studies with high statistical power.
Therefore, the primary objectives of this article were to determine and (1) to investigate potential patterns of association between perioperative circulating NT-proBNP levels and early outcomes after repair of congenital heart defects and (2) to evaluate the reliability of NT-proBNP level as a predictor for postoperative outcomes in a large series of children undergoing cardiac surgery for CHD.
Patients and Methods
The clinical information system databases of the Guangzhou Women and Children's Medical Center were scanned for pediatric patients enrolled for cardiac surgery having CHD. Patients younger than 18 years with CHD who underwent cardiac surgery were eligible for the study. Children who had preoperative arrhythmia (potentially malignant ventricular arrhythmias and high-degree atrioventricular block), rheumatic heart disease, infective endocarditis, myocarditis, pericardial disease, renal dysfunction, or neoplasms,
or who could not be separated from cardiopulmonary bypass (CPB) were excluded.
Between June and December 2014, 415 eligible consecutive patients undergoing corrective or palliative CHD surgery at the Department of Pediatric Cardiac Surgery and admitted to the Cardiac Intensive Care Unit (CICU) at the Guangzhou Women and Children's Medical Center, Guangzhou, were identified. However, to display the results in a more homogeneous population, we excluded 38 cases without CPB and 14 single-ventricular cases from the final analysis according to the reviewers. Thus, results from 363 cases were presented in the current study. The institutional review board at the Guangzhou Women and Children's Medical Center reviewed and approved this study.
The preoperative anesthesia management, surgical procedure (including sternotomy, establishing CPB, correction of cardiac defects, closure of thoracic cavity, and suturing the skin; Video 1), and subsequent CICU management followed standard institutional practices. All patients were admitted to the CICU intubated and mechanically ventilated after operation. If postoperative urine output was less than 0.5 mL/kg per hour lasting for more than 2 to approximately 4 hours under the premise of the application of diuretics, peritoneal dialysis was adopted until the urine output was more than 2 mL/kg per hour, blood urea nitrogen, creatinine, and electrolytes were maintained in the normal range. Termination of the surgical procedure means ending with suturing the skin.
Sample and Data Collection
Blood samples were obtained from an arterial catheter preoperatively and at 1, 12, and 36 hours after the surgical procedure at CICU according to our routine clinical protocol. The NT-proBNP levels were measured with a commercially available fluorescence immunoassay (competitive Enzyme Immuno Assay; ReLIA II, Shenzhen, China) and Multi-Detection Microplate Reader (VICTOR X5; PerkinElmer, Waltham, Mass). The measurable range of NT-proBNP on this device is between 5 and approximately 35,000 pg/mL (range of normality according to age and gender).
Clinical and biochemical data were collected retrospectively from the medical records. The clinical data collected included patient demographics (age, weight, gender, main diagnosis), the Risk Adjustment in Congenital Heart Surgery, version 1 (RACHS-1) score, CPB duration, aorta cross clamp (ACC) time, and follow-up measurements after surgical procedure, including duration of mechanical ventilation, intensive care duration, inotropic drug duration, mean systemic arterial pressure, central venous pressure, serum creatinine, and urine output. The patients were divided into 3 subgroups of age according to clinically relevant data (neonates <30 days, infants up to 1 year, and 1 to 12 years).
Categorical data were expressed as percentages and compared by using the Fisher exact test or the χ2 test. Continuous variables were expressed as mean and standard deviation. They were assessed for normality by the Kolmogorov-Smirnov test and compared by using the Student t test or by the 1-way analysis of variance for 2- or multiple-group comparisons. Because of their abnormal distribution, NT-proBNP levels and the primary predictors (including the duration of mechanical ventilation, the length of ICU stay, and the duration of inotropic therapy) were all log-transformed for analytical purposes. Given repeated measures of NT-proBNP at fixed time points before and after surgery, we fit linear mixed effects model, combining fixed and random subject effects, for log(NT-proBNP) levels over time.
Moreover, we were also interested in the association between perioperation NT-proBNP changes over time and the duration of mechanical ventilation, ICU stay, and inotropic therapy. To analyze these data, we propose a Bayesian joint model with (1) a linear mixed effects (LME) model for the longitudinal analysis of repeated NT-proBNP measurements, and (2) a generalized linear model for the continuous primary outcomes. We used the Markov chain Monte Carlo (MCMC) technique via Gibbs sampler to estimate the unknown parameters simultaneously using publically available WinBUGS software. See Chen et al
for detailed discussions of the Bayesian modeling approach and the implementation of the MCMC procedures.
Receiver operating characteristic (ROC) curves were used to assess the various cutoff values of NT-proBNP to predict the need for mechanical ventilation beyond 48 hours, ICU stay beyond 3 days, and duration of inotropic therapy more than 72 hours. The areas under the curve (AUCs) for the NT-proBNP measured at different time points were compared by using De Long's method. Next, NT-proBNP levels were dichotomized to being equal to or less than the cutoff values. All tests were performed 2-tailed, with P < .05 considered statistically significant. Statistical analysis was performed using SPSS version 19.0 (IBM SPSS Statistics for Windows; IBM Corporation, Somers, NY) and R software (version 3.2.2).
The patients’ clinical characteristics are listed in Table 1. There was no mortality among the enrolled patients during the study period. A total of 363 patients experienced CPB, excluding those diagnosed of patent ductus arteriosus, coarctation, and the single-ventricular group that accepted palliative operations. Ninety patients (24.79%) had cyanotic CHD.
Table 1Clinical characteristics of cases undergoing congenital heart disease surgery
Type of congenital heart disease
Ventricular septal defect
Atrial septal defect
Total anomalous pulmonary venous connection
Tetralogy of Fallot
Transposition of the great arteries
Complete atrioventricular septal defect
Double outlet right ventricle
Interrupted aortic arch
Other types of CHD
15.8 ± 27.4
Body weight, kg
7.4 ± 5.7
NT-proBNP level, pg/mL
1-h postoperative level
12-h postoperative level
36-h postoperative level
CPB time, min
Aortic crossclamp time, min
Mechanical ventilation, h
Intensive care unit stay, d
Inotropic duration, d
Data are expressed as n (%) or median with range. CHD, Children with congenital heart disease; RACHS-1, risk Adjustment for Congenital Heart Surgery-1.
The level of postoperative NT-proBNP showed a rising trend regardless of preoperative condition. Mean levels of NT-proBNP increased significantly to the peak at 12 hours to 36 hours after CPB or surgical procedure.
The fixed effects estimates from the longitudinal LME model for change in log(NT-proBNP) are presented in Table 2. NT-proBNP levels were found to have significantly relative decreases of 76.8% and 74.1% at preoperative and 1-hour postoperative time points, but an increase of 47.7% at 12 hours after surgery, compared with that measured at 36 hours after surgery. Neonates and infants were found to have 124.8% and 76.8% higher expected NT-proBNP than the group of older children. Children who had RACHS-1 score ≤2 or with deep hypothermic circulatory arrest (DHCA) were found to have 31.6% and 56.0% lower expected NT-proBNP than the score ≥3 or patients without DHCA, respectively. Postoperative creatinine level and weight were also significant predictors of NT-proBNP levels. However, no differences in expected NT-proBNP were found between male and female patients, or with the change of the length of ACC and CPB time; respective P values were .16, .10, and .36.
Table 2Estimates for linear mixed model of repeated NT-proBNP
As shown in Table 3, the results for the longitudinal submodel were consistent with the results from the separate LME analysis. The differences in magnitudes of the parameter estimates were negligible and there was no material difference in terms of statistical significance (Tables 2 and 3). Hence, in what follows, we discuss only the results for the second submodels. Most importantly, we found that a 1% increase in NT-proBNP level was associated with a relative increase of 5.5% in expected duration of the mechanical ventilation. Compared with 36-hour postoperative NT-proBNP level, 1-hour (P = .03) and 12-hour postoperative NT-proBNP (P = .01) were also significantly associated with the duration of mechanical ventilation.
Table 3Results for the joint modeling analysis of a linear mixed effects model for NT-proBNP from before to 36 hours after surgery and a generalized linear model for the duration of the mechanical ventilation
All patients were discharged between 1 and 41 days. Time-varying NT-proBNP was associated with the prolonged ICU stay. A 1% increase in NT-proBNP was associated with a 3.9% prolonged duration of ICU stay. Patients whose RACHS-1 score was ≤2 were found to have 8.6% lower expected ICU stay than patients with a score ≥3. Patients who experienced DHCA were estimated to have a 10.3% higher expected ICU stay than patients who did not experience DHCA (Table 4).
Table 4Results for the joint modeling analysis of a linear mixed effects model for NT-proBNP from before to 36 hours after surgery and a generalized linear model for the duration of intensive care unit stay
On joint model analysis, NT-proBNP and RACHS-1 score were found to be independent predictors of prolonged duration of inotropic therapy (Table 5). Having adjusted for confounding variables, there was still a 3.5% (multivariate P = .01) increase in the likelihood of prolonged duration of inotropic therapy for each 1% increase in NT-proBNP.
Table 5Results for the joint modeling analysis of a linear mixed effects model for NT-proBNP from before to 36 hours after surgery and a generalized linear model for the duration of inotropic therapy
On ROC curve analyses, NT-proBNP measured at 1 hour after surgery had better performance as compared with the 36-hour postoperative level in predicting longer duration of mechanical ventilation (>48 hours) (AUC, 0.84 vs 0.76; z = 2.08; P = .04), as shown in Figure 1, Figure 2, Figure 3, Figure 4. However, no significant differences in discrimination between the perioperative NT-proBNP levels at different time points for both prolonged ICU stay (>3 days) and duration of inotropic therapy (>3 days) were found. The optimal cut points for preoperative and 1-hour, 12-hour, and 36-hour postoperative NT-proBNP and the corresponding sensitivity and specificity in detecting longer duration of mechanical ventilation, ICU stay, and inotropic therapy are shown in Figure 1.
The current study suggests and expands some already present views that perioperative NT-proBNP levels in the children undergoing CHD constructive surgery are predictive of postoperative affairs (eg, the prolonged mechanical ventilation time, ICU stay, and inotropic support duration after surgery).
The prognostic values of perioperative NT-proBNP levels in the setting of cardiac surgery have been mainly determined in the adult population and to a lesser extent in pediatrics. As far as we know, this is the largest study to assess the prognostic value of NT-proBNP in unselected heterogeneously aged children undergoing cardiac surgery for CHD. Furthermore, this is also the first study reporting the prognostic utility of NT-proBNP level at as early as 1 hour after surgery in this population, suggesting an important role for its use in postoperative management.
This study also characterized the alternations of NT-proBNP levels during the perioperative period of children undergoing surgical repair for CHD. Postoperative NT-proBNP values begin to increase 1 hour after surgery, reaching a maximum level at 12 hours after surgery. However, Pérez-Piaya et al
observed the NT-proBNP values start to rise 6 hours after surgery, peaking at 24 hours after the operation in a prospective study of 68 children (0-15 years) submitted to open-heart surgery. The inconsistencies among studies may result from a variety of reasons; for example, different sampling time, different bypass or myocardial protecting techniques, or different postoperative medical management.
Given the difference of BNP/NT-proBNP levels in children according to age,
we also evaluated the alterations in NT-proBNP levels of children and neonates over time in separate analyses. Neonates (n = 30), however, on average show relatively lower levels at 1 hour after surgery (before surgery: 12,131 ± 10,123 pg/mL → 1 hour after surgery: 9610 ± 5592 pg/mL → 12 hours after surgery: 13,438 ± 9711 pg/mL → 36 hours after surgery: 12,037 ± 6298 pg/mL). Similar results were reported by Cannesson et al
in neonates (n = 30) undergoing transposition of the great arteries, of which BNP level progressively declined after surgery compared with the baseline and then increased gradually over the next 24 hours. This different postsurgery pattern of BNP/NT-proBNP kinetics compared with older children may in part be contributed to the generally higher disease severity and greater operation complexity.
Among perioperative NT-proBNP values at 4 different time points (from before to 1, 12, and 36 hours after surgery), all were found to be positively correlated with duration of mechanical ventilation, ICU stay, and inotropic drug therapy. Value at 1 hour after operation had the highest prediction value (Figure 1). Thus, why are 1-hour postoperative NT-proBNP levels more strongly correlated with subsequent prognoses compared with the preoperative value? It may be that BNP/NT-proBNP values after cardiac surgery reflect the comprehensive cardio-renal function and the extension of surgical response.
Thus, BNP/NT-proBNP levels measured either before or at 12 hours after surgery will prevent prognostic utility among these patients. To our delight, the 1-hour NT-proBNP level after surgery (Figure 1) was also predictive of prolonged mechanical ventilation, ICU stay, and inotropic support treatment; that is, this earlier time point might also have clinical utility within the first 12 to 48 hours postoperatively.
Furthermore, it is well known that circulating levels of BNP/NT-proBNP in patients can be affected by various extracardiac conditions. Thus, it needs to be highlighted that postoperative NT-proBNP levels and their subsequent prognostic value may be affected by several factors. First, given different NT-proBNP response at 1-hour after surgery between neonates and older children, the predictive values of this index have been evaluated in separate analysis. In neonates, results showed that with the exception of the relationship between preoperative levels and ICU stay, no significant association between perioperative NT-proBNP levels and mechanical ventilation, ICU stay, or inotropic support treatment duration were found. Similarly, a study about BNP by Hsu et al
Besides the current study, we also evaluated it in a homogeneous cohort without CPB (n = 38). It is confirmed that results of patients without CPB were consistent and in line with the results of cases with CPB.
Another major factor is with regard to the assessment of renal function, which could confound the blood NT-proBNP levels.
In our study, children were excluded from the analysis if they had evidence of impaired renal function before surgery. Moreover, acute kidney injury (AKI) (defined mainly based on the urine output for early identification) was recorded in 15 cases. Subgroup analysis showed no significant difference of NT-proBNP levels before and after surgery in patients with and without AKI. This may be related to our timely intervention to the potential AKI patients using peritoneal dialysis.
There are several limitations of the present study that may provide further extension of the research. First, we did not use Doppler imaging for the assessment of ventricular systolic and diastolic function, which may add further prognostic information. Compared with the RACHS-1 scoring system, the Society of Thoracic Surgery-European Association for Cardio-Thoracic Surgery (STAT) for surgical complexity is a more widely accepted risk score in the current era. STAT was not routinely used in pediatric cardiac surgery in our center until now. We cannot adjust STAT as a confounder in the multivariable regression models. However, one recent study
comparing stratification of complexity models (RACHS-1, Aristotle basic score, and STAT) in patients younger than 18 years with CHD, showed that 3 models are with similar discriminatory capacity for hospital mortality. Compared with comprehensive use of both RACHS-1 model and STAT model, the RACHS-1 single model used in our center has some limitations. Obviously, there was also certain rationality based on the results of the comparison of the 2 models. Obviously, we will try to introduce the STAT system in our clinical setting to keep with modern clinical practices. Second, prognostic cutoff points from the ROC analyses are specific for the mentioned population of children without impaired renal function before surgery. The results are specific and can be recommended for use in homogeneous populations. Finally, our study was also limited by the sample size and was not powered to perform subgroup analyses to evaluate the effect of age (eg, neonates), type of CHD, and kinds of reconstruction performed. There is insufficient power to draw definitive conclusions in these subgroups.
In conclusion, perioperative NT-proBNP levels in the children undergoing CHD surgery, particularly levels measured 1 hour after surgery, provide stronger prognostic information and might be a more early, accurate, and sensitive predictor of prolonged ventilation, ICU stay, and inotropic support duration. Further studies are warranted to confirm if the routine NT-proBNP monitoring should be performed during postoperative management to identify individuals at high risk of adverse events who might benefit from alternative treatment strategies.
Conflict of Interest Statement
Authors have nothing to disclose with regard to commercial support.
The authors thank the doctors and nurses at the Department of Heart Center, Guangzhou Women and Children's Medical Center, in particular Xinxin Chen, MD, and his team of surgeons. We thank Ramit Kumar Gupta, MD, for assisting with the English language and preparing the manuscript, and Dr Taishan Zeng from the School of Mathematical Sciences, South China Normal University, for his statistical support, particularly for the joint model using.
The whole procedure of the artery switch operation (ASO) for transposition of great artery (TGA) treatment. This kind of surgical procedure is one of the typical surgeries for congenital heart disease undergoing cardiopulmonary bypass. Video available at: http://www.jtcvsonline.org/article/S0022-5223(17)30188-5/addons.
Casey Jr., D.E.
2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Pratice Guidelines.
In a recent article, Qu and colleagues1 reported their observations on the relationship between N-terminal pro–brain natriuretic peptide (NT-proBNP) levels at different time points and early outcome in an attempt to evaluate the reliability of NT-proBNP level as a predictor of early outcome after congenital heart surgery. It would certainly be useful to have such a tool available in the perioperative management of children with congenital heart defects undergoing surgery, but there is probably space for improvement in the proposed methodology.