We sought to determine prospectively whether serial assessment of the natriuretic peptide prohormone, amino-terminal pro–B-type natriuretic peptide (NT–pro-BNP), correlated with clinical severity and outcomes in children hospitalized for acute decompensated heart failure (ADHF). Patients (>1 month of age) admitted from 2005 to 2007 with ADHF requiring intravenous vasoactive/diuretic therapy for ADHF were eligible. Serum NT–pro-BNP levels were obtained within 24 hours of admission and at prespecified intervals, and clinical caregivers were blinded to these levels. End points included hospital discharge, death or cardiac transplantation, and care escalation including the need for mechanical circulatory support (MCS) was noted. Twenty-four patients were enrolled: 22 survived to hospital discharge and 2 died. Ten required MCS (of which 6 underwent cardiac transplantation). Two patients underwent transplantation without MCS. For the entire cohort, NT–pro-BNP levels peaked at days 2 to 3 after admission, with a subsequent gradual decrease until discharge. However, for those who did require MCS, NT–pro-BNP failed to decrease consistently until after MCS initiation. At discharge, NT–pro-BNP levels were significantly decreased from admission levels but remained well above normal for all patients. Single-point NT–pro-BNP levels on admission did not correlate with independently assessed clinical scores of heart failure severity or predict the need for MCS in this cohort. In conclusion, serial NT–pro-BNP levels demonstrated an incremental trend after 48 hours in patients who went on to require MCS but decreased in all other patients and may therefore assist the decision to initiate or avoid MCS after admission for pediatric ADHF.
In children, acute decompensated heart failure (ADHF) due to cardiomyopathy or failed congenital heart disease repair constitutes a common indication for cardiac transplantation. Admission with severe HF frequently raises concerns that a patient has declared the need for transplantation assessment. In this regard, recent advances have made use of mechanical circulatory support (MCS) in pediatrics a more popular and feasible method of bridging to transplantation or to recovery. However, the appropriate timing of MCS device placement is sometimes difficult to determine. We investigated whether initial assessment of amino-terminal pro–B-type natriuretic peptide (NT–pro-BNP) in children admitted for management of ADHF correlated with formal assessment of clinical status, and whether serial assessment of NT–pro-BNP might be a useful tool to predict the need for MCS in children with ADHF of diverse causes.
Methods
This was a prospective observational study conducted with the approval of the Hospital for Sick Children (Toronto, Ontario, Canada) research ethics board. Patients admitted to this institution from May 2005 to July 2007 with a diagnosis of ADHF requiring escalation of HF management involving a need for intravenous diuretics, inotropic medication, or MCS were eligible for enrollment. Diagnosis of HF was made by the clinician responsible for admission and not the research team. Two of the investigators (D.W. and P.K.) adjudicated each case for eligibility based on specific criteria before enrollment. Inclusion criteria were (1) a clinical diagnosis of ADHF, (2) age >1 month to <18 years at admission, and (3) escalation of HF therapy with intravenous diuretic therapy, inotropic medication, or MCS. Patients with congenital or acquired heart disease were equally eligible. Exclusion criteria included (1) planned corrective cardiovascular surgery or catheter-based intervention on the same admission (to avoid the effect of surgical or catheter intervention as a confounder of patient outcome) and (2) patients <1 month of age because their NT–pro-BNP levels are known to be increased as part of the normal postnatal circulatory adaptation. All patients who met the criteria for enrollment were approached and informed consent was obtained.
On admission, demographic information, underlying cardiac diagnosis, laboratory investigations (i.e., biochemistry, radiologic studies, electrocardiograms), and echocardiographic data were recorded by the research team. A brief symptomatic history and physical examination were performed at time of admission and at time of discharge by members of the research team (D.W., J.W., and K.G.). A patient’s clinical course in hospital—length of stay, discharge status and medications—and need for inotropic support, diuretics, vasoactive medications, antiarrhythmics, and anticoagulation were recorded. We noted whether the outcome of each admission was 1 of the predefined end points of discharge, death, or heart transplantation. Other clinically relevant events were also defined and noted prospectively, including CCCU admission, mechanical ventilation, and inotropic and vasoactive medication usage. At time of admission and at time of attaining a defined end point, the previously validated New York University Pediatric Heart Failure Index (NYU-PHFI) score, Ross classification, and where age appropriate the New York Heart Association (NYHA) score were determined. To isolate the effect of the NT–pro-BNP level from clinical severity assessment and the decision to escalate hemodynamic support, clinicians and the research team were blinded to the NT–pro-BNP data until the study was completed.
NT–pro-BNP levels were obtained within 24 hours of admission, at intervals of 2 to 3, 6 to 8, and 13 to 15 days after admission, and weekly thereafter. Patients who were admitted to the critical care unit had NT–pro-BNP levels drawn daily. If patients were on stable long-term MCS for >1 week, NT–pro-BNP levels were drawn weekly while in the critical care unit. A final sample was drawn within 48 hours of planned discharge or immediately before heart transplantation. Samples were stored in clotted blood specimen containers at −80°C and analyzed after completion of the entire study by our clinical laboratory as a research protocol. Assays were performed on an Elecsys electro-chemiluminescent immunoassay system using Elecsys 1010/2010 immunoassay analyzers (Roche Diagnostics, Laval, Quebec, Quebec, Canada) with appropriate calibration and control methods. Because neither NT–pro-BNP nor BNP levels were routinely measured in our institution at the time of this study, no patient received any clinically indicated BNP or NT–pro-BNP assay during admission. Investigators were also blinded to results of the NT–pro-BNP assays until after the clinical component of the study was completed.
Data are presented as mean ± SD, medians with minimum and maximum values, and frequencies as appropriate. NT–pro-BNP levels over time are represented as box-plots to express the severely skewed level distribution. To account for the skewed distribution of NT–pro-BNP, a natural logarithmic transformation was applied to NT–pro-BNP in all analyses. Basic comparisons between end point groups (MCS vs no MCS) were obtained through Fisher’s exact test, Student’s t test with Satterthwaite correction, and Kruskal-Wallis analysis of variance for continuous variables with skewed distribution. Factors associated with need for MCS were sought in univariable logistic regression models using need for MCS before discharge as a binary variable. No multivariable modeling was attempted due to the limited number of patients enrolled in the study. Changes in NT–pro-BNP over time and associated factors were assessed in linear regression models adjusted for repeated measurements over time to an autoregressive (first order with model-based estimation of covariance structure). Regression parameters were estimated using generalized estimating equations. Mean values were used for imputation of missing variables when necessary. Time since admission in days was treated as a continuous variable. An interaction criterion was created between study groups (MCS vs not) and time to estimate differences in rate of change in NT–pro-BNP for each group. Because of the small number of patients who died or required heart transplantation we were not able to search for predictors of negative outcomes (even as a combined end point). All statistical analyses were performed using SAS 9.1 (SAS Institute, Cary, North Carolina).
Results
The clinical pathway overview for the cohort is diagrammed in Figure 1 .
In all, 24 consecutive patients consented to enrollment. Median age (range) was 8.7 years (0 to 17.7) and median weight was 29 kg (3 to 80). Cardiomyopathy was the underlying cause of HF in 17 cases, including idiopathic (8), myocarditis (2), ischemic (1), arrhythmia-induced (2), and unclassified (4) phenotypes. Structural congenital heart disease was present in 4 patients, with 3 other patients including 1 with post-transplant rejection and hemodynamic failure.
Eighteen patients were in Ross/NYHA class III or IV at admission, with 5 in class II. One patient (an infant with a mitochondrial cardiomyopathy) appeared clinically asymptomatic to the research team but later deteriorated precipitously, requiring MCS. Most patients showed a change in symptomatic status between admission and the end point, with all those dying or requiring transplantation reaching stage NHYA/Ross class IV before that end point. Median ejection and shortening fractions, as measured by echocardiography on admission for the entire cohort, were 23% (range 6 to 64) and 14% (range 3 to 50), respectively. Only 7 patients were treated with diuretic therapy alone, followed by initiation of oral angiotensin-converting enzyme inhibiter and/or β blocker. Most patients in this cohort (17 of 24) required inotropic or inodilator medications—17 received milrinone, 9 received epinephrine, 6 received vasopressin, 4 received norepinephrine, and 3 received dobutamine. Eleven patients required a combination of multiple inotropic and/or vasoactive medications. No patients received nesiritide, which was unavailable in Canada at the time of this study. Three patients received digoxin during their hospitalization.
Twelve patients (50%) were admitted to the critical care unit, and all of these required mechanical ventilation. Ten patients required MCS, of which 7 were managed with extracorporeal membranous oxygenation and 3 patients required the Berlin Heart EXCOR (Berlin Heart, Berlin, Germany) ventricular assist device. Differences in clinical presentation characteristics between those patients who went on to require MCS and those who did not are presented in Table 1 . Patients who eventually required MCS had a significantly lower pH and a higher lactate and creatinine on admission than those patients who did not require MCS. In the group requiring MCS median time between admission and initiation of support was 13 days (range 0 to 36). Mean duration on MCS was 26 ± 33 days (median 17, range 2 to 107). Eight patients underwent heart transplantation, 6 of whom underwent bridging by MCS before their heart transplantation. Two patients died—1 patient immediately after heart transplantation (after needing MCS for failed single-ventricle palliation) and the other patient having care withdrawn after a large intracranial hemorrhage while on MCS for HF due to dilated cardiomyopathy. Overall survival to hospital discharge was 92% for this cohort, with a transplant-free survival to discharge of 63%.
| Patient Characteristics | MCS | No MCS | p Value |
|---|---|---|---|
| (n = 10) | (n = 14) | ||
| New York University Pediatric Heart Failure Index | 13 (1–24) | 13 (9–21) | NS |
| Ross/New York Heart Association classification | 3.5 (2–4) | 3 (1–4) | 0.06 |
| Lactate (mmol/L) | 7.8 (3.3–12.6) | 2.0 (1.2–5.8) | 0.02 |
| pH | 7.25 (7.03–7.41) | 7.41 (7.28–7.46) | 0.01 |
| Creatinine (μmol/L) | 89 (31–150) | 46 (4–83) | 0.02 |
| Left ventricular ejection fraction (%) | 17 (9–41) | 30 (9–64) | 0.05 |
| Amino-terminal pro–B-type natriuretic peptide (pg/ml), median (range) | |||
| On admission | 30,406 (11,400–63,862) | 10,743 (3,760–62,221) | 0.03 |
| 2–3 days | 39,653 (20,793–87,899) | 9,262 (3,094–70,580) | 0.01 |
| 6–8 days | 26,510 (3,529–39,252) | 8,058 (568–63,847) | 0.03 |
| 13–15 days | 16,245 (3,204–43,740) | 8,594 (1,070–32,775) | 0.12 |
| At discharge/exit | 3,984 (766–48,397) | 8,933 (3,612–38,270) | 0.56 |
| Maximum | 40,713 (25,204–87,899) | 15,680 (7,111–70,580) | 0.001 |
Serum NT–pro-BNP levels were available for all patients. Duration of hospitalization varied significantly depending on the clinical course of each patient (range 2 to 125 days), and as a result the number of NT–pro-BNP measurements per patient also varied. Sixteen patients had measurements that spanned all 5 specified intervals. The trend of NT–pro-BNP over time for the entire cohort is displayed in Figure 2 . It can be appreciated that the level of NT–pro-BNP was 100 to 1,000 times the upper limit of normal (normal range <200 ng/ml). Despite a carefully conducted clinical severity assessment, we found that the correlation coefficient between HF symptom scores at admission and NT–pro-BNP levels was weak ( Figure 3 ). This applied to the NYHA/Ross class status (r = 0.15, p = 0.52) and the NYU-PHFI (r = 0.17, p = 0.46).
