Abstract
Rationale: Brain-type natriuretic peptide (BNP) correlates with pulmonary hypertension as demonstrated by echocardiogram in congenital diaphragmatic hernia (CDH); however, its association with right ventricular (RV) function and mortality is unknown.
Objectives: To characterize the relationships between echocardiogram-derived RV strain, BNP, and mortality in diaphragmatic hernia.
Methods: We performed a single-center retrospective cohort study of infants with CDH and at least one BNP–echocardiogram pair within a 24-hour period. RV global longitudinal strain (GLS) and free-wall strain (FWS) were measured on existing echocardiograms. Associations among strain, BNP, and mortality were tested using mixed-effect linear and logistic regression models. Survival analysis was stratified by BNP and strain abnormalities.
Results: There were 220 infants with 460 BNP–echocardiogram pairs obtained preoperatively (n = 237), ≤1 week postoperatively (n = 35), and >1 week postoperatively (“recovery”; n = 188). Strain improved after repair (P < 0.0001 for all periods). Higher BNP level was associated with worse strain in recovery but not before or immediately after operation (estimate [95% confidence interval] for recovery: GLS, 1.03 [0.50–1.57]; P = 0.0003; FWS, 0.62 [0.01–1.22]; P = 0.047). BNP and strain abnormalities were associated with an extracorporeal-membrane oxygenation requirement. Higher BNP level in recovery was associated with greater mortality (odds ratio, 11.2 [1.2–571.3]; P = 0.02). Abnormal strain in recovery had high sensitivity for detection of mortality (100% for GLS; 100% for FWS) but had low specificity for detection of mortality (28% for GLS; 48% for FWS).
Conclusions: Persistent RV dysfunction after CDH repair may be detected by a high BNP level and abnormal RV strain.
Keywords: congenital diaphragmatic hernias, pulmonary hypertension, heart, ventricles, pediatrics
Neonates and infants with pulmonary hypoplasia due to congenital diaphragmatic hernia (CDH) may experience life-threatening pulmonary hypertension and right ventricular (RV) failure. Maldevelopment of the pulmonary vasculature results in elevated pulmonary vascular resistance and pulmonary artery pressure (1), which are poorly tolerated by the right ventricle. Despite surgical advances, CDH morbidity and mortality are high and have been attributed to pulmonary hypoplasia and pulmonary hypertension (2, 3).
In CDH, treatment of pulmonary hypertension and RV failure may be guided by the echocardiogram and brain-type natriuretic peptide (BNP), a peptide secreted by the cardiac myocytes in response to pressure overload (4). BNP is a well-established marker of ventricular strain in adults with heart failure (5) and pulmonary hypertension (6) and correlates with the presence of pulmonary hypertension in CDH (7). In adults with pulmonary arterial hypertension (PAH), BNP increases with RV dysfunction and is associated with mortality (8, 9). However, the relationships among BNP, quantitative measures of RV function, and CDH mortality are unknown.
Qualitative ventricular dysfunction by standard two-dimensional echocardiographic imaging was a critical determinant of mortality in a CDH registry–based study (10). Quantitative assessment of RV function is challenging because of the right ventricle’s pyramidal geometry, anterior position in the chest, and altered geometry in disease states (11, 12). More recently, myocardial deformation (“strain”) imaging has emerged as a sensitive and accurate geometry-independent method to identify RV dysfunction (13, 14). Strain refers to the change in length of the myocardium relative to its resting length (expressed as a percentage, with more negative values indicating better systolic function). Reference ranges of RV global longitudinal strain (GLS) in healthy children are available (15). In children and adults with PAH, better RV GLS predicts survival (16, 17). In neonates with CDH, RV strain is decreased after delivery (18–20), and worse RV strain is associated with the need for extracorporeal-membrane oxygenation (ECMO) (21).
In clinical practice, an elevated BNP level may suggest worsening pulmonary hypertension and/or RV dysfunction; however, the relationship between BNP and RV strain has not been established in patients with CDH. Therefore, the objectives of this study were to characterize RV GLS and free-wall strain (FWS) in neonates and infants with CDH and to investigate the associations among RV strain, BNP, and mortality.
Methods
Study Cohort
This single-center, retrospective cohort study included all infants ≤6 months of age with CDH enrolled in the Children’s Hospital of Philadelphia Pulmonary Hypoplasia Program between 2005 and 2017 who had at least one echocardiogram and serum BNP test performed within 24 hours of each other. Infants with structural heart disease, other than small atrial or ventricular septal defects, were excluded, as were those with persistent large patent ductus arteriosus with left-to-right shunt. According to our laboratory’s standard protocol, a large patent ductus arteriosus is defined as having a measure equal to or larger than the branch pulmonary arteries. Echocardiograms performed while weaning ECMO flow and low-quality images on which strain analyses could not be performed were excluded.
On the basis of the number of patients with CDH enrolled in our Pulmonary Hypoplasia Program database, we estimated that approximately 220 eligible infants would have at least two BNP–echocardiogram pairs per infant, resulting in at least 440 pairs of data. With this sample size, we would have 81% power to detect a partial correlation coefficient of 0.135 between BNP level and RV GLS or FWS with a 5% significance level using multiple linear regression adjusting for other covariates. The study was approved by the Children’s Hospital of Philadelphia Institutional Review Board.
Clinical Variables
BNP levels and pertinent clinical data were abstracted from the electronic medical record. Quantitative plasma measurement of BNP was performed using a chemiluminescent microparticle immunoassay on an Abbot Architect i2000SR. BNP was considered abnormal if greater than 100 pg/ml, the upper limit of normal for our laboratory.
Echocardiographic Variables
Clinically indicated echocardiograms were performed using standard pediatric views with 3- to 8-MHz transducers on a Phillips IE33 machine in accordance with our echocardiography laboratory’s standard imaging protocol, which was consistent over the study period. Images were digitally stored in the Syngo Dynamics system (Siemens). A single pediatric cardiac sonographer (Y.W.) blinded to clinical characteristics obtained offline measures of RV function. Strain measurements were obtained from speckle-tracking analysis of four-chamber cine-loop images including the RV free wall and ventricular septum using Tomtec software (Image Arena 4.6). Values were averaged to calculate RV GLS and FWS (percentages). By convention, strain values are presented as negative numbers, with a higher magnitude of absolute values indicating better systolic function. For example, a strain value of −30% reflects better systolic function than a strain value of −20%. Strain was considered abnormal if it was less than the absolute value of −25% (15). RV systolic function was also assessed by M-mode tricuspid annular plane systolic excursion (TAPSE), fractional area change (FAC), and tricuspid annular peak systolic velocity (s’) measured by tissue doppler (22). TAPSE z-scores (TAPSEZ) were generated from normative data (23, 24). The FAC (percentage) was calculated from the apical four-chamber view as the end-diastolic area minus the end-systolic area divided by the end-diastolic area (25). Pulmonary hypertension was considered present if at least one echocardiographic criterium was identified: 1) RV pressure greater than one-half the systemic systolic blood pressure estimated from tricuspid regurgitant jet or patent ductus arteriosus (when present) velocity by the modified Bernoulli Equation (4 × velocity2) without an estimate of right atrial pressure, 2) bidirectional or right-to-left patent ductus arteriosus (or rarely ventricular septal defect) shunt, or 3) flattened or bowing ventricular septum at end-systole (26). We have previously published high intrarater (for sonographer Y.W.) and interrater intraclass correlation coefficient values of 0.9–0.99 for RV strain and other continuous echocardiographic measures of RV function (27–30).
Statistical Analyses
First, we described the demographic and clinical characteristics of the study cohort. Next, we compared the BNP, strain, and other echocardiographic measurements among different time periods. Then, we assessed the association between BNP level and strain overall and within each time period. Finally, we investigated the relationship between BNP and strain abnormalities and survival. Several subanalyses are also described. A P value < 0.05 was used to define statistical significance.
Cohort Characteristics
Standard descriptive statistics were used to describe infant characteristics, BNP levels, and echocardiographic parameters. Echocardiographic and BNP data were grouped relative to date of CDH repair: preoperative (“pre-op”) data (before CDH repair), immediate postoperative (“post-op”) data (≤1 wk after CDH repair), and “recovery” data (>1 wk after CDH repair). Another category, “after repair,” included both post-op and recovery data. In our clinical experience, many patients have improved hemodynamics and are weaning from cardiorespiratory support about 1 week after CDH repair, so we differentiated the immediate post-op period (≤1 wk after CDH repair) from the later recovery period.
Differences in BNP and Echocardiographic Measures across Time Periods
Kruskal-Wallis tests were used to compare median BNP level and continuous echocardiographic variables among the three time periods using average values for patients with more than one BNP–echocardiogram pair in a given time period. The presence of pulmonary hypertension among the three time periods was compared using mixed-effects logistic regression models.
Relationship between Strain and Other Markers of RV Function, Overall and within Each Time Period
Linear mixed-effect models were used to test the association between strain and other markers of RV systolic function as continuous variables, with strain as the dependent variable. These models included days after surgical repair in post-op regression models.
Relationship between BNP and Strain, Overall and within Each Time Period
Generalized linear mixed-effect models were used to determine the association between BNP level and strain as continuous variables. Logarithmic transformation of BNP was performed. Simple linear mixed models were used to assess the relationship between demographic/clinical variables and strain. Variables included sex, race, ethnicity, gestational age, birth weight, CDH side, presence of liver herniation, history of ECMO, patch repair, age at CDH repair, and days after surgical repair. The variables that were statistically significant at a 0.05 level were adjusted as covariates in the BNP and strain models and included history of ECMO, days after surgical repair, and age at CDH repair. History of ECMO was adjusted as a covariate in the models in all three time periods. Days after surgical repair and age at CDH repair were adjusted as covariates in the post-op regression models. Sensitivity analyses were conducted via stratified models by ECMO requirement. Estimates (95% confidence intervals) were calculated for the regression models in each time period.
Relationship between BNP and Strain Abnormalities and Survival
Logistic regressions were used to test the association between the presence of BNP or strain abnormality and survival. After using simple logistic regression models to test other factors that could affect survival, ECMO and CDH repair age were included in the model as covariates. The sensitivity, specificity, positive predictive value, and negative predictive value of BNP and strain abnormalities in detecting mortality were calculated.
Additional Subanalyses
We also analyzed existing, clinically indicated echocardiograms performed on any patient from the original cohort at 6 ± 1 months of age. Strain and other echocardiographic measures were performed as above. BNP levels obtained within 1 week of the echocardiogram were extracted from the medical record. The associations between BNP level and strain and between strain and other measures of RV systolic function were tested by Spearman correlation. The Wilcoxon rank-sum test was used to compare strain in infants with and without echocardiographic evidence of pulmonary hypertension.
Results
There were 220 infants who met inclusion criteria, resulting in 460 BNP–echocardiogram pairs (range, 1–6 pair(s)/subject). Infant characteristics are shown in Table 1. Most patients were male, born at term, and had left CDH and liver herniation. ECMO was required in 69 (31%) patients, and 33 (7%) BNP–echocardiogram pairs were obtained in subjects while on ECMO. CDH repair occurred at median 13 (interquartile range [IQR], 6–22) days of age and included a patch in more than half of subjects. There were 29 deaths (13%) with 14 (48%) occurring before repair and 15 (52%) occurring at a median of 34 (IQR 7.5–91) post-op days.
Table 1.
Infant characteristics (N = 220)
| n (%) or Median (IQR) | |
|---|---|
| Sex | |
| Male | 128 (58) |
| Female | 92 (42) |
| Race | |
| White | 172 (78) |
| Black | 7 (7) |
| Asian | 13 (6) |
| Biracial | 10 (5) |
| Other | 9 (4) |
| Ethnicity | |
| Hispanic or Latino | 38 (17) |
| Gestational age, wk | 38.4 (37 to 39) |
| Birthweight, g | 3,060 (2,750 to 3,450) |
| Left sided CDH | 184 (83) |
| Liver herniation | 119 (54) |
| ECMO support | 69 (31) |
| Patch CDH repair | 130 (63) |
| Age at CDH repair, d | 13 (6–22) |
| Death | 29 (13) |
Definition of abbreviations: CDH = congenital diaphragmatic hernia; ECMO = extracorporeal-membrane oxygenation; IQR = interquartile range.
There were 237 (52%) BNP–echocardiogram pairs obtained before CDH repair, 35 (8%) pairs obtained ≤1 week immediately after operation, and 188 (40%) pairs obtained in recovery >1 week after operation. Table 2 and Figures 1 and 2 demonstrate the change in BNP, RV strain, and other echocardiographic parameters with CDH repair. Abnormal BNP was seen in 89% in the pre-op period, 91% in the immediate post-op period, and in 42% in recovery. The presence of abnormal GLS and FWS strain was distributed as follows: pre-op period = 86% and 82%, immediate post-op period = 85% and 76%, and recovery period = 43% and 55%, respectively. BNP, GLS, and FWS all improved after CDH repair (P < 0.0001 for all comparisons among time periods).
Table 2.
Improvement in BNP, RV strain, and other markers of RV function with CDH repair
| Variable | Overall (n = 460) | Pre-Op (n = 237) | After Repair* (n = 223) | Immediate Post-Op (n = 35) | Recovery (n = 188) | P Value† |
|---|---|---|---|---|---|---|
| BNP | 231 (65.1 to 671.1) | 360.7 (186.9 to 923.6) | 92.9 (23.7 to 409.6) | 498.7 (175.3 to 1,415.6) | 62 (20.7 to 282.3) | <0.0001 |
| RV GLS | −20 (−15.8 to −23.9) | −17.9 (−14.8 to −22.2) | −21.5 (−17.9 to −25) | −19.3 (−15.3 to −23.8) | −21.8 (−18.5 to −25.4) | <0.0001 |
| RV FWS | −22.6 (−18.5 to −27) | −20.5 (−16.5 to −24.4) | −24.6 (−21.1 to −28.5) | −21.9 (−17.9 to −25.3) | −25 (−21.8 to −28.8) | <0.0001 |
| RV FAC, % | 37.1 (30.5 to 44.8) | 34 (26.8 to 40.8) | 40.7 (34.2 to 48) | 35.4 (30.1 to 42.9) | 41.1 (35.2 to 48.5) | <0.0001 |
| TAPSE, cm‡ | 0.9 (0.7 to 1.1) | 0.7 (0.6 to 0.9) | 1.1 (0.9 to 1.3) | 0.9 (0.8 to 1.15) | 1.1 (0.9 to 1.3) | <0.0001 |
| TAPSEZ | −0.9 (−2.7 to 0.1) | −1.8 (−3.5 to 0.9) | −0.3 (−1.4 to 1.1) | −0.09 (−1 to 1.6) | −0.3 (−1.4 to 1.1) | <0.0001 |
| RV s’, m/s§ | 0.07 (0.06 to 0.09) | 0.06 (0.05 to 0.08) | 0.09 (0.07 to 0.1) | 0.08 (0.06 to 0.09) | 0.09 (0.08 to 0.1) | <0.0001 |
| Pulmonary hypertension | 225 (95) | 225 (95) | 118 (53) | 28 (80) | 9 (48) | <0.0001 |
Definition of abbreviations: BNP = brain-type natriuretic peptide; CDH = congenital diaphragmatic hernia; FAC = fractional area change; FWS = free-wall strain; GLS = global longitudinal strain; post-op = postoperative; pre-op = preoperative; RV = right ventricular; s’ = tricuspid annular peak systolic velocity; TAPSE = tricuspid annular plane systolic excursion; TAPSEZ = TAPSE z-score.
Continuous variables are expressed as the median (interquartile range). Categorical variables are expressed as n (%).
Includes both immediate post-op and recovery data.
P value for comparisons among the pre-op, immediate post-op, and recovery periods.
For TAPSE/TAPSEZ: overall n = 294, pre-op n = 147, post-op n = 147, immediate post-op n = 20, recovery n = 127.
For RV s’: overall n = 269, pre-op n = 134, post-op n = 135, immediate post-op n = 18, recovery n = 117.
Figure 1.
BNP decreases after congenital diaphragmatic hernia repair. The P value is for comparisons among the pre-op, immediate post-op, and recovery periods. BNP = brain-type natriuretic peptide; post-op = postoperative; pre-op = preoperaptive.
Figure 2.
Both (A) GLS and (B) FWS improve after congenital diaphragmatic hernia repair. P value is for comparisons among the pre-op, immediate post-op, and recovery periods. FWS = free-wall strain; GLS = global longitudinal strain; post-op = postoperative; pre-op = preoperaptive.
Worse GLS and FWS were associated with other markers of RV dysfunction. Both were associated with lower FAC in all time periods and with lower TAPSEZ in the pre-op and recovery periods (Table 3). Strain was not consistently associated with RV s’ velocity; GLS was associated with RV s’ velocity only in recovery, and FWS was associated with RV s’ velocity only in the pre-op period. Strain was worse in those with pulmonary hypertension in recovery but was not worse in the pre-op or immediate post-op periods.
Table 3.
Association between strain and echocardiographic measures of RV function by time period
| Dependent Variable | Predictor | Estimate (95% CI); P Value* |
||||
|---|---|---|---|---|---|---|
| All Data | Pre-Op | After Repair† | Immediate Post-Op | Recovery | ||
| GLS | TAPSEZ | −0.88 (−1.19 to −0.58); <0.0001 | −0.89 (−1.46 to 0.31); 0.0033 | −0.58 (−0.99 to −0.17); 0.006 | −0.29 (−1.60 to 1.02); 0.64 | −0.70 (−1.14 to −0.26); 0.0024 |
| RV FAC, % | −0.48 (−0.51 to −0.45); <0.0001 | −0.53 (−0.57 to −0.49); <0.0001 | −0.42 (−0.47 to −0.37); <0.0001 | −0.52 (−0.63 to −0.40); <0.0001 | −0.39 (−0.45 to −0.34); <0.0001 | |
| RV s’, m/s | −0.85 (−1.19 to −0.51); <0.0001 | −0.38 (−1.04 to −0.27); 0.24 | −0.58 (−1.09 to −0.08); 0.026 | 0.20 (−2.65 to 3.04); 0.89 | −0.79 (−1.30 to −0.28); 0.0036 | |
| Pulmonary hypertension | 4.71 (3.37 to 6.04); <0.0001 | 2.30 (−3.38 to 7.99); 0.32 | 3.45 (1.90 to 5.00); 0.0001 | −4.08 (−10.05 to 1.89); 0.17 | 3.47 (1.85 to 5.08); 0.0003 | |
| FWS | TAPSEZ | −0.94 (−1.27 to −0.61); <0.0001 | −1.01 (−1.60 to −0.42); 0.0013 | −0.46 (−0.92 to −0.01); 0.046 | −0.21 (−1.43 to 1.02); 0.72 | −0.56 (−1.08 to −0.05); 0.03 |
| RV FAC, % | −0.48 (−0.52 to −0.45); <0.0001 | −0.52 (−0.57 to −0.48); <0.0001 | −0.40 (−0.46 to −0.34); <0.0001 | −0.48 (−0.60 to −0.36); <0.0001 | −0.38 (−0.45 to −0.30); <0.0001 | |
| RV s’, m/s | −0.85 (−1.20 to −0.49); <0.0001 | −0.76 (−1.40 to −0.13); 0.021 | −0.08 (−0.66 to −0.50); 0.78 | 0.09 (−2.69 to 2.88); 0.94 | −0.16 (−0.78 to 0.46); 0.61 | |
| Pulmonary hypertension | 5.05 (3.64 to 6.46); <0.0001 | 2.43 (−3.37 to 8.23); 0.31 | 3.10 (1.41 to 4.80); 0.0009 | −3.86 (−9.12 to 2.41); 0.24 | 3.07 (1.24 to 4.91); 0.0026 | |
Definition of abbreviations: 95% CI = 95% confidence interval; FAC = fractional area change; FWS = free-wall strain; GLS = global longitudinal strain; post-op = postoperative; pre-op = preoperative; RV = right ventricle; s’ = tricuspid annular peak systolic velocity; TAPSEZ = tricuspid annular plane systolic excursion z-score.
All P values are for comparisons among pre-op, immediate post-op, and recovery periods.
Includes both immediate post-op and recovery data.
BNP and strain were positively associated such that as BNP level increased, strain also increased (lower magnitude of absolute value), which corresponded to a decrease in RV systolic function (Table 4). However, this relationship depended on the time period relative to CDH repair. There was no association between BNP level and strain in either the pre-op or immediate post-op periods, but there was a positive association between BNP level and strain in the recovery period. Higher BNP level was associated with worse strain in recovery (estimates: GLS, 1.03 [0.50–1.57]; P = 0.0003; FWS, 0.62 [0.01–1.22]; P = 0.047) but not before or after operation.
Table 4.
Association between BNP (independent) and strain (dependent variable)
| Dependent Variable | Estimate [ β Coefficient (95% CI); P Value]* |
||||
|---|---|---|---|---|---|
| All Data (n = 460) | Pre-Op (n = 237) | After Repair† (n = 223) | Immediate Post-Op (n = 35) | Recovery (n = 188) | |
| GLS | 1.18 (0.82 to 1.55); <0.0001 | 0.39 (−0.38 to 1.16); 0.32 | 1.00 (0.50 to 1.49); 0.0001 | 1.61 (−0.49 to 3.72); 0.13 | 1.03 (0.50 to 1.57); 0.0003 |
| FWS | 1.27 (0.89 to 1.66); <0.0001 | 0.50 (−0.28 to 1.29); 0.21 | 0.73 (0.18 to 1.27); 0.0094 | 1.25 (−0.81 to 3.31); 0.22 | 0.62 (0.01 to 1.22); 0.047 |
Definition of abbreviations: 95% CI = 95% confidence interval; BNP = brain-type natriuretic peptide; FWS = free-wall strain GLS = global longitudinal strain; post-op = postoperative; pre-op = preoperative.
All P values are for comparisons among pre-op, immediate post-op, and recovery periods.
Includes both immediate post-op and recovery data.
When the relationships between patient characteristics and BNP/strain abnormalities were tested, a history of ECMO was associated with abnormal BNP and strain after CDH repair (Table 5). No other patient characteristic was associated with abnormal BNP or strain. A subanalysis comparing patients with and without a history of ECMO demonstrated a persistent positive association between recovery BNP level and strain in the patients with a history of ECMO (Table 6). BNP level and strain were still positively associated in patients without a history of ECMO, but the association only trended toward significance when examined by time period.
Table 5.
History of ECMO was associated with abnormal BNP/strain after CDH repair
| Dependent Variable (after Repair) | Predictor | Odds Ratio (95% CI) | P Value |
|---|---|---|---|
| BNP | ECMO | 2.27 (1.05–4.76) | 0.038 |
| GLS | ECMO | 2.86 (1.27–6.25) | 0.013 |
| FWS | ECMO | 4.17 (2.13–7.69) | <0.0001 |
Definition of abbreviations: 95% CI = 95% confidence interval; BNP = brain-type natriuretic peptide; CDH = congenital diaphragmatic hernia; ECMO = extracorporeal-membrane oxygenation; FWS = free-wall strain; GLS = global longitudinal strain.
Table 6.
Association between BNP (independent variable) and strain (dependent variable) for patients with ECMO history (n = 69 with 152 BNP-echo pairs) and without ECMO history (n = 151 with 308 BNP–echo pairs)
| Dependent Variable | Estimate [β Coefficient (95% CI); P Value]* |
||||
|---|---|---|---|---|---|
| All Data | Pre-Op | After Repair† | Immediate Post-Op | Recovery | |
| Patients with ECMO history |
|||||
| n = 152 | n = 74 | n = 78 | n = 12 | n = 66 | |
| GLS | 1.33 (0.64 to 2.03); 0.0002 | 0.98 (−0.41 to 2.36); 0.16 | 1.61 (0.79 to 2.44); 0.0004 | 1.92 (−1.71 to 5.55); 0.27 | 1.52 (0.66 to 2.37); 0.0012 |
| FWS | 1.16 (0.43 to 1.89); 0.0022 | 1.21 (−0.21 to 2.63); 0.09 | 0.94 (0.10 to 1.78); 0.029 | 0.99 (−2.51 to 4.50); 0.54 | 0.90 (0.03 to 1.78); 0.043 |
| Patients without ECMO history |
|||||
| n = 308 | n = 163 | n = 145 | n = 23 | n = 122 | |
| GLS | 1.12 (0.68 to 1.55); <0.0001 | 0.10 (−0.84 to 1.03); 0.84 | 0.52 (−0.11 to 1.15); 0.10 | 1.36 (−1.52 to 4.24); 0.34 | 0.50 (−0.21 to 1.20); 0.16 |
| FWS | 1.33 (0.87 to 1.79); <0.0001 | 0.16 (−0.80 to 1.13); 0.74 | 0.49 (−0.22 to 1.20); 0.17 | 1.46 (−1.37 to 4.29); 0.30 | 0.23 (−0.60 to 1.06); 0.57 |
Definition of abbreviations: 95% CI = 95% confidence interval; BNP = brain-type natriuretic peptide; echo = echocardiogram; ECMO = extracorporeal-membrane oxygenation; FWS = free-wall strain; GLS = global longitudinal strain; post-op = postoperative; pre-op = preoperative.
All P values are for comparisons among pre-op, immediate post-op, and recovery.
Includes both immediate post-op and recovery data.
Abnormal BNP after CDH repair was associated with mortality (post-op odds ratio [OR], 12.2 [1.4–638.0]; P = 0.02; recovery OR, 11.2 [1.2–571.3]; P = 0.03). All deceased patients with available data had abnormal GLS and FWS after CDH repair (Table 7). There was a trend toward significance between abnormal FWS and mortality in the recovery period (P = 0.097). In recovery, both GLS and FWS abnormalities were highly sensitive but poorly specific in detecting mortality (GLS: sensitivity of 100%, specificity of 28%, positive predictive value of 8%, and negative predictive value of 100%; FWS: sensitivity of 100%, specificity of 48%, positive predictive value of 11%, and negative predictive value of 100%).
Table 7.
Strain abnormalities in deceased patients versus survivors
| Abnormal GLS | Abnormal FWS | |
|---|---|---|
| After repair* | ||
| Deceased, n = 10 | 10 (100%) | 10 (100%) |
| Alive, n = 134 | 103 (77%) | 71 (57%) |
| P value, exact logistic regression | 0.29 | 0.12 |
| Recovery | ||
| Deceased, n = 8 | 8 (100%) | 8 (100%) |
| Alive, n = 122 | 88 (72%) | 63 (52%) |
| P value, exact logistic regression | 0.29 | 0.097 |
Definition of abbreviations: FWS = free-wall strain; GLS = global longitudinal strain.
Includes patients with abnormal strain in either immediate postoperative or recovery periods.
Of the 220 patients, 81 (37%) had an echocardiogram at 6 ± 1 months of age. In these, 29 of 81 (36%) had evidence of pulmonary hypertension and 63 of 81 (78%) had a BNP level. There was no correlation between the median BNP level (18.7 [11.6 to 37.5]; range, 10 to 1,930) and strain (GLS: −26.2 [−22.1 to −29.5]; FWS −28.9 [−25.2 to −32.3]) at 6 months (P = 0.11 for GLS; P = 0.24 for FWS). GLS was correlated with FAC (Spearman r = −0.64; P < 0.0001), and FWS was correlated with both TAPSEZ (r = −0.34; P = 0.02) and FAC (r = −0.68; P < 0.0001). Strain was worse in patients with pulmonary hypertension (GLS in pulmonary hypertension: −25% [−19 to −27] vs. GLS in nonpulmonary hypertension: −27% [−23 to −30]; P = 0.018; FWS in pulmonary hypertension: −27% [−24 to −31] vs. FWS in nonpulmonary hypertension: −30% [−26 to −33]; P = 0.045).
Discussion
In this large cohort of neonates and infants with CDH, we demonstrated low initial RV strain with improvement in strain and BNP level after CDH repair. In recovery from CDH repair, higher BNP level was associated with worse RV strain and with mortality. Abnormal RV strain detected mortality with high sensitivity but low specificity. To our knowledge, this is the largest report of RV strain in patients with CDH to date and the first to investigate the relationships among strain, BNP, and mortality. These findings increase our appreciation of the scope of RV dysfunction in CDH and increase our knowledge of strain as a prognostic tool in this population.
Several small studies have described abnormal strain in newborns with CDH. RV strain is decreased compared with age-matched control subjects (18, 20) and worse in patients with CDH requiring ECMO than in patients not requiring ECMO (21). In addition, a decreased magnitude of left ventricular (LV) GLS, but not RV strain, in the first 48 hours of life is associated with ECMO and death (19). Similar to authors of other studies, we also found that post-op BNP and strain abnormalities were associated with a history of ECMO (21). However, our study is the first to describe changes in RV strain with CDH repair and beyond the immediate neonatal period. It is concerning that abnormal strain was still seen in approximately half of the cohort after repair. These new data are essential to recognizing persistent RV dysfunction in CDH to develop targeted pharmacologic interventions to improve cardiac function and patient outcomes.
BNP and nonactive N-terminal pro-BNP are strong prognostic indicators in adult and pediatric PAH (8, 9, 31). Although BNP correlates with the presence of pulmonary hypertension in CDH, it does not discriminate pulmonary hypertension severity (7). Our findings demonstrate that the relationship between BNP and RV systolic function may vary on the basis of CDH repair status. The lack of association between BNP level and RV strain in the pre-op period may not be surprising. Pre-op cardiac care may include inotropes, lusitropic agents, pulmonary vasodilators, and prostaglandins to maintain ductus arteriosus patency in the face of severe pulmonary hypertension or RV dysfunction, which may result in differences in RV “load” and BNP level among otherwise similar patients. The small number of BNP–echocardiogram pairs in the immediate post-op period may have affected the lack of association in that period, however, after 1 week of post-op recovery, there was a significant association between BNP level and RV strain. In our experience, many patients are weaning from cardiorespiratory support after 1 post-op week, and BNP level may then be a more accurate reflection of RV load. The associations among BNP, strain, and mortality in recovery may reflect the utility of BNP as a longer-term marker of RV function and predictor of mortality.
Our findings suggest that close monitoring for RV dysfunction after CDH repair is warranted and that management of RV dysfunction could improve patient outcomes. Members of our group previously reported early and sustained improvement in RV GLS in 47 children with pulmonary hypertension from various causes, including some with CDH, after initiation of treprostinil (27). In that study, BNP level also improved with treatment, although the relationship between BNP and strain was not investigated. The effects of vasodilator therapies on RV strain and patient outcomes in CDH should be further investigated to determine whether the treatment effect can be measured by improvement in RV strain.
Finally, 36% of patients who had undergone an echocardiogram at 6 months had evidence of pulmonary hypertension. BNP levels were generally normal, although this may have reflected pulmonary vasodilator treatment. The lack of correlation between BNP level and RV strain at 6 months may be due to the lack of variability in BNP. However, RV strain was worse in patients with pulmonary hypertension than in those without pulmonary hypertension. The prognostic implications of this difference warrant further evaluation, but this is consistent with prior reports of decreased RV strain several years after CDH repair (32). These findings support the need for continued surveillance of ventricular function in these patients.
Limitations
Our study is limited by its retrospective nature. We only included patients with high-quality images on which strain could be measured retrospectively. We included patients with BNP–echocardiogram pairs performed within 24 hours of each other, so we cannot exclude management changes in response to one measure that would have affected the other. The scope of the study did not allow us to assess the effect of management strategies on strain or the BNP–strain relationship, but this should be tested in future prospective studies. Most of our analyses were not significant during the immediate post-op period. This could be due to the small number of BNP–echocardiogram pairs in that time period. Finally, LV systolic dysfunction is an important prognostic indicator in CDH. Future studies should describe LV strain over time and investigate the association among BNP, LV strain, and outcomes.
Conclusions
In conclusion, RV dysfunction, as reflected by a lower magnitude of RV strain, improves with CDH repair but persists in many patients. Higher BNP level is associated with worse RV strain and with mortality during recovery from CDH repair. Strain abnormalities are associated with mortality with high sensitivity; therefore, normal strain values are reassuring. BNP and strain should be followed closely after CDH repair, and appropriate management of RV dysfunction is warranted. Future interventions targeting RV dysfunction in this population may help improve survival in patients with CDH.
Supplementary Material
Footnotes
Supported by a Children’s Hospital of Philadelphia Cardiac Center grant (C.M.A.), U.S. National Institutes of Health grant K01HL125521 (L.M.-R.), and a Pulmonary Hypertension Association supplement to K01HL125521 (L.M.-R.).
Author Contributions: C.M.A. and L.M.-R. conceived of and designed the work; acquired, analyzed, and interpreted data; approved the final version; and agree to be accountable for all aspects of the work. C.M.A. drafted the manuscript. All authors revised the manuscript. Y.W. acquired and analyzed all echocardiographic data. X.Z., S.S., S.A., L.H., D.B.F., M.D.Q., N.E.R., H.L.H., and L.M.-R. made substantial contributions to the conception of the work, revised the manuscript, approved the final version, and agree to be accountable for all aspects of the work.
Author disclosures are available with the text of this article at www.atsjournals.org.
References
- 1.Rottier R, Tibboel D. Fetal lung and diaphragm development in congenital diaphragmatic hernia. Semin Perinatol. 2005;29:86–93. doi: 10.1053/j.semperi.2005.04.004. [DOI] [PubMed] [Google Scholar]
- 2.Chinoy MR. Pulmonary hypoplasia and congenital diaphragmatic hernia: advances in the pathogenetics and regulation of lung development. J Surg Res. 2002;106:209–223. doi: 10.1006/jsre.2002.6390. [DOI] [PubMed] [Google Scholar]
- 3.Chao PH, Huang CB, Liu CA, Chung MY, Chen CC, Chen FS, et al. Congenital diaphragmatic hernia in the neonatal period: review of 21 years’ experience. Pediatr Neonatol. 2010;51:97–102. doi: 10.1016/S1875-9572(10)60018-6. [DOI] [PubMed] [Google Scholar]
- 4.Hunt PJ, Yandle TG, Nicholls MG, Richards AM, Espiner EA. The amino-terminal portion of pro-brain natriuretic peptide (pro-BNP) circulates in human plasma. Biochem Biophys Res Commun. 1995;214:1175–1183. doi: 10.1006/bbrc.1995.2410. [DOI] [PubMed] [Google Scholar]
- 5.Abassi Z, Karram T, Ellaham S, Winaver J, Hoffman A. Implications of the natriuretic peptide system in the pathogenesis of heart failure: diagnostic and therapeutic importance. Pharmacol Ther. 2004;102:223–241. doi: 10.1016/j.pharmthera.2004.04.004. [DOI] [PubMed] [Google Scholar]
- 6.Boucly A, Weatherald J, Savale L, Jaïs X, Cottin V, Prevot G, et al. Risk assessment, prognosis and guideline implementation in pulmonary arterial hypertension. Eur Respir J. 2017;50:1700889. doi: 10.1183/13993003.00889-2017. [DOI] [PubMed] [Google Scholar]
- 7.Partridge EA, Hanna BD, Rintoul NE, Herkert L, Flake AW, Adzick NS, et al. Brain-type natriuretic peptide levels correlate with pulmonary hypertension and requirement for extracorporeal membrane oxygenation in congenital diaphragmatic hernia. J Pediatr Surg. 2015;50:263–266. doi: 10.1016/j.jpedsurg.2014.11.009. [DOI] [PubMed] [Google Scholar]
- 8.Nagaya N, Nishikimi T, Okano Y, Uematsu M, Satoh T, Kyotani S, et al. Plasma brain natriuretic peptide levels increase in proportion to the extent of right ventricular dysfunction in pulmonary hypertension. J Am Coll Cardiol. 1998;31:202–208. doi: 10.1016/s0735-1097(97)00452-x. [DOI] [PubMed] [Google Scholar]
- 9.Nagaya N, Nishikimi T, Uematsu M, Satoh T, Kyotani S, Sakamaki F, et al. Plasma brain natriuretic peptide as a prognostic indicator in patients with primary pulmonary hypertension. Circulation. 2000;102:865–870. doi: 10.1161/01.cir.102.8.865. [DOI] [PubMed] [Google Scholar]
- 10.Patel N, Lally PA, Kipfmueller F, Massolo AC, Luco M, Van Meurs KP, et al. Ventricular dysfunction is a critical determinant of mortality in congenital diaphragmatic hernia. Am J Respir Crit Care Med. 2019;200:1522–1530. doi: 10.1164/rccm.201904-0731OC. [DOI] [PubMed] [Google Scholar]
- 11.Ho SY, Nihoyannopoulos P. Anatomy, echocardiography, and normal right ventricular dimensions. Heart. 2006;92:i2–i13. doi: 10.1136/hrt.2005.077875. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sheehan F, Redington A. The right ventricle: anatomy, physiology and clinical imaging. Heart. 2008;94:1510–1515. doi: 10.1136/hrt.2007.132779. [DOI] [PubMed] [Google Scholar]
- 13.Valsangiacomo Buechel ER, Mertens LL. Imaging the right heart: the use of integrated multimodality imaging. Eur Heart J. 2012;33:949–960. doi: 10.1093/eurheartj/ehr490. [DOI] [PubMed] [Google Scholar]
- 14.DiLorenzo MP, Bhatt SM, Mercer-Rosa L. How best to assess right ventricular function by echocardiography. Cardiol Young. 2015;25:1473–1481. doi: 10.1017/S1047951115002255. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Levy PT, Sanchez Mejia AA, Machefsky A, Fowler S, Holland MR, Singh GK. Normal ranges of right ventricular systolic and diastolic strain measures in children: a systematic review and meta-analysis. J Am Soc Echocardiogr. 2014;27:549–560, e3. doi: 10.1016/j.echo.2014.01.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Okumura K, Humpl T, Dragulescu A, Mertens L, Friedberg MK. Longitudinal assessment of right ventricular myocardial strain in relation to transplant-free survival in children with idiopathic pulmonary hypertension. J Am Soc Echocardiogr. 2014;27:1344–1351. doi: 10.1016/j.echo.2014.09.002. [DOI] [PubMed] [Google Scholar]
- 17.Park JH, Park MM, Farha S, Sharp J, Lundgrin E, Comhair S, et al. Impaired global right ventricular longitudinal strain predicts long-term adverse outcomes in patients with pulmonary arterial hypertension. J Cardiovasc Ultrasound. 2015;23:91–99. doi: 10.4250/jcu.2015.23.2.91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Massolo AC, Paria A, Hunter L, Finlay E, Davis CF, Patel N. Ventricular dysfunction, interdependence, and mechanical dispersion in newborn infants with congenital diaphragmatic hernia. Neonatology. 2019;116:68–75. doi: 10.1159/000499347. [DOI] [PubMed] [Google Scholar]
- 19.Patel N, Massolo AC, Paria A, Stenhouse EJ, Hunter L, Finlay E, et al. Early postnatal ventricular dysfunction is associated with disease severity in patients with congenital diaphragmatic hernia. J Pediatr. 2018;203:400–407, e1. doi: 10.1016/j.jpeds.2018.07.062. [DOI] [PubMed] [Google Scholar]
- 20.Altit G, Bhombal S, Van Meurs K, Tacy TA. Diminished cardiac performance and left ventricular dimensions in neonates with congenital diaphragmatic hernia. Pediatr Cardiol. 2018;39:993–1000. doi: 10.1007/s00246-018-1850-7. [DOI] [PubMed] [Google Scholar]
- 21.Altit G, Bhombal S, Van Meurs K, Tacy TA. Ventricular performance is associated with need for extracorporeal membrane oxygenation in newborns with congenital diaphragmatic hernia. J Pediatr. 2017;191:28–34, e1. doi: 10.1016/j.jpeds.2017.08.060. [DOI] [PubMed] [Google Scholar]
- 22.Saxena N, Rajagopalan N, Edelman K, López-Candales A. Tricuspid annular systolic velocity: a useful measurement in determining right ventricular systolic function regardless of pulmonary artery pressures. Echocardiography. 2006;23:750–755. doi: 10.1111/j.1540-8175.2006.00305.x. [DOI] [PubMed] [Google Scholar]
- 23.Koestenberger M, Ravekes W, Everett AD, Stueger HP, Heinzl B, Gamillscheg A, et al. Right ventricular function in infants, children and adolescents: reference values of the tricuspid annular plane systolic excursion (TAPSE) in 640 healthy patients and calculation of z score values. J Am Soc Echocardiogr. 2009;22:715–719. doi: 10.1016/j.echo.2009.03.026. [DOI] [PubMed] [Google Scholar]
- 24.Koestenberger M, Nagel B, Ravekes W, Urlesberger B, Raith W, Avian A, et al. Systolic right ventricular function in preterm and term neonates: reference values of the tricuspid annular plane systolic excursion (TAPSE) in 258 patients and calculation of Z-score values. Neonatology. 2011;100:85–92. doi: 10.1159/000322006. [DOI] [PubMed] [Google Scholar]
- 25.Koestenberger M, Friedberg MK, Nestaas E, Michel-Behnke I, Hansmann G. Transthoracic echocardiography in the evaluation of pediatric pulmonary hypertension and ventricular dysfunction. Pulm Circ. 2016;6:15–29. doi: 10.1086/685051. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Mourani PM, Mandell EW, Meier M, Younoszai A, Brinton JT, Wagner BD, et al. Early pulmonary vascular disease in preterm infants is associated with late respiratory outcomes in childhood. Am J Respir Crit Care Med. 2019;199:1020–1027. doi: 10.1164/rccm.201803-0428OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hopper RK, Wang Y, DeMatteo V, Santo A, Kawut SM, Elci OU, et al. Right ventricular function mirrors clinical improvement with use of prostacyclin analogues in pediatric pulmonary hypertension. Pulm Circ. 2018;8:2045894018759247. doi: 10.1177/2045894018759247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.DiLorenzo MP, Elci OU, Wang Y, Banerjee A, Sato T, Ky B, et al. Longitudinal changes in right ventricular function in tetralogy of Fallot in the initial years after surgical repair. J Am Soc Echocardiogr. 2018;31:816–821. doi: 10.1016/j.echo.2018.02.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Himebauch AS, Yehya N, Wang Y, Conlon T, Kilbaugh TJ, McGowan FX, et al. Early right ventricular systolic dysfunction and pulmonary hypertension are associated with worse outcomes in pediatric acute respiratory distress syndrome. Crit Care Med. 2018;46:e1055–e1062. doi: 10.1097/CCM.0000000000003358. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Himebauch AS, Yehya N, Wang Y, McGowan FX, Mercer-Rosa L. New or persistent right ventricular systolic dysfunction is associated with worse outcomes in pediatric acute respiratory distress syndrome. Pediatr Crit Care Med. 2020;21:e121–e128. doi: 10.1097/PCC.0000000000002206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Ploegstra MJ, Zijlstra WM, Douwes JM, Hillege HL, Berger RM. Prognostic factors in pediatric pulmonary arterial hypertension: a systematic review and meta-analysis. Int J Cardiol. 2015;184:198–207. doi: 10.1016/j.ijcard.2015.01.038. [DOI] [PubMed] [Google Scholar]
- 32.Egan MJ, Husain N, Stines JR, Moiduddin N, Stein MA, Nelin LD, et al. Mid-term differences in right ventricular function in patients with congenital diaphragmatic hernia compared with controls. World J Pediatr. 2012;8:350–354. doi: 10.1007/s12519-012-0380-2. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.