Patent ductus arteriosus, systemic NT-proBNP concentrations and development of bronchopulmonary dysplasia in very preterm infants: retrospective data analysis from a randomized controlled trial

Patent ductus arteriosus (PDA) occurs in more than 50% of preterm infants < 28 weeks of gestation [1]. It is known that there is an association between PDA and the risk of bronchopulmonary dysplasia (BPD) developing and/or death before 36 weeks of postmenstrual age (PMA). However, the risk does not depend on either presence or absence of PDA, but on the magnitude of PDA shunt [2, 3]. A large left-to-right PDA shunt leads to increased pulmonary blood flow and alters the pulmonary fluid filtration rate, which can cause pulmonary edema, increased pulmonary pressure, and induce structural changes that may contribute to the development of BPD [4].

A recent study showed that infants < 28 weeks of gestation who were exposed to moderate-to-large PDA shunts for at least 7–13 days had a significantly increased risk of developing BPD or death, and exposure for at least 14–27 days affected the BPD risk alone [5]. However, it is unknown whether this association is causal, as the results of numerous randomized controlled trials using prophylactic or early PDA treatment have not shown a decrease in the incidence of BPD or death [6, 7]. Age at the time of PDA treatment did not significantly affect its effectiveness either [8]. With increased survival rates achieved in recent decades, BPD is diagnosed in about 25% of very-low-birth-weight infants [9]. BPD is a common potentially dangerous disease that worsens the long-term prognosis for health and development [9]. Due to the lack of effective treatment, prevention of this disease is important and it should begin as soon as possible. Therefore, it is very important to predict possible development of BPD in very preterm newborns. The most important risk factor for BPD is gestational age, but almost a third of extremely preterm infants do survive without the disease [10]. The search for additional prognostic criteria that would help to solve this problem is still ongoing. Among such criteria, important attention is paid to biomarkers pathogenetically related to the development of BPD, the advantage of which is objectivity and feasibility. However, the diagnostic value of many tested biomarkers remains insufficient, thus mandating the need for reliable early indicators of BPD development [11].

A recent study by Rodríguez-Blanco et al. showed that N-terminal pro-brain natriuretic peptide (NT-proBNP) level in very preterm infants may be a much more informative indicator for the prediction of BPD than echocardiography, because the elevated levels of NT-proBNP more accurately predicted BPD development than the detection of hsPDA by echocardiography alone [12]. The systematic review and meta-analysis of studies [13] showed that the sensitivity and specificity of NT-proBNP levels in preterm infants ≤32 weeks of gestation in the prediction of BPD or death ranged from 61 to 85% and from 71 to 73%, respectively, and in the prediction of BPD only – from 89 to 100% and from 69 to 86%, respectively. However, the infants’ age at the time of NT-proBNP testing was different in different studies, ranging from the first 48 h of life until PMA of 36 weeks. The included studies differed in methodology, peptide determination methods, prognostic thresholds, gestational age, etc.

We hypothesized that serum NT-proBNP levels may be elevated in very preterm infants at risk of developing BPD/death in the first 10 days after birth, and this association is independent of the presence of PDA. Therefore, our study aimed to evaluate the relationship between PDA, early serum NT-proBNP levels, and BPD/death in very preterm infants < 32 weeks of gestation.

hsPDA is a common complication in very preterm infants born at < 32 weeks of gestation. It is known that a large left-to-right PDA shunt significantly increases the risk of developing BPD or death [2, 3]. The results of this study showed that infants who died or developed BPD significantly more often had hsPDA than babies who survived without BPD. Our data confirmed the results of a recent meta-analysis [13], according to which in five out of nine studies the hsPDA or PDA was significantly more likely diagnosed in infants who developed BPD. According to our results, death or BPD occurred significantly more often in infants with lower gestational age, birth weight, and the condition of such infants after birth was more compromised, as evidenced by significantly lower Apgar scores, the need for intubation and MV in delivery room as well as by the significantly higher SNAPPE-II scores on the first day of NICU hospitalization.

The median size of PDA in the first 3 days of life was significantly smaller in infants who survived without BPD. Echocardiographic markers of hsPDA at this age were significantly more often seen in infants who died or developed BPD. At the same time, according to our results, pharmacological treatment of hsPDA did not significantly reduce the incidence of BPD. Therefore, despite the association between PDA and BPD/death in epidemiological studies, screening and early treatment of infants with PDA in recent years have not reduced the risk of BPD, suggesting that the relationship between PDA and BPD may not be causal [20]. In particular, the association between hsPDA and severe neonatal morbidity differs in different studies, remaining one of the most controversial aspects of the problem [21, 22].

It is now known that there is a dose-dependent correlation between the magnitude of the PDA shunt and the risk of BPD. El-Khuffash et al. [23] proposed a PDA score based on echocardiography measurements at 48 h after birth (n = 141). Their results showed that gestational age, together with echocardiographic markers of PDA and left ventricular diastolic function, may predict the later occurrence of BPD or death in preterm infants < 29 weeks of gestation. A PDA score cut-off of 5 or more had sensitivity and specificity of 92 and 87%, and the positive and negative predictive values of 92 and 82%, respectively [23]. Thus, the relationship between PDA and BPD strengthens with decreasing gestational age and increasing ductal flow. At the same time, the duration of PDA exposure can also play an important role in the development of BPD. Schena et al. [2] assessed PDA significance using staging system proposed by McNamara and Sehgal [24] in infants < 29 weeks of gestation (n = 242). According to their results, the risk of developing BPD increased by 70% for each additional week of exposure to the hsPDA. A recent study by Mirza et al. [25] also showed that the longer duration of hsPDA in extremely preterm infants was associated with a higher risk of death or BPD development (aOR = 1.37; 95% CI 1.03–1.82). However, our results did not prove the association between the duration of hsPDA and the risk of death or BPD development.

One of the clinical predictors of BPD development is a pattern of oxygen dependence in the first 2 weeks of life [26]. The level of oxygen dependency in our patients on the eighth day of life was not associated with later BPD development, possibly due to the predominance of mild BPD in our patients. At the same time, the results of our study showed that serum NT-proBNP values at 8–9 days of life were significantly associated with FiO₂ at this age.

The results of lung ultrasound have high diagnostic value in predicting the risk of developing BPD, but only after 1 week of life [27]. Despite the diagnostic value of echocardiography and lung ultrasound, the interpretation of the results of both these examinations is subjective, requires involvement of skilled specialists, and the examination itself requires additional costs. Determination of biomarkers allows to eliminate the subjectivity, does not require additional equipment and specialists, and thus could be cheaper and more practical. The relationship between hsPDA and the risk of developing BPD is known, but the mechanism of this association remains poorly understood and requires further investigation.

The prognostic value of potential BPD biomarkers has not been sufficiently studied, as the feasible and reliable biochemical indicator that could accurately predict the development of BPD in the first days of life in very preterm infants has not yet been identified [10]. It is known that systemic NT-proBNP concentrations have a high sensitivity and specificity for predicting the hsPDA in the first days of life [11]. Serum NT-proBNP concentrations ≥17,745 pg/ml in our patients at a median age of 2 days had 55% sensitivity and 81% specificity for predicting BPD or death. Our data are consistent with the results of a study by Sellmeer & Singer [17], which showed that the serum NT-proBNP concentrations on the third day of life could reliably predict the development of BPD/death in infants < 32 weeks of gestation (AUC = 0.76; 95% CI: 0.68–0.84). At the same time, their results, like ours, showed that in infants with hsPDA who later developed BPD or died, the serum NT-proBNP levels at 3 days of life were significantly higher than in infants with hemodynamically insignificant or closed ductus [17].

According to the results of our study, the median NT-proBNP value at 8–9 days of life was not only associated with PDA persistency up to 10 days of life but also reliably determined the risk of BPD or death. Serum NT-proBNP concentrations ≥3537 pg/ml at 8–9 days of life had 60% sensitivity and 89% specificity for the prediction of BPD or death regardless of the PDA persistency till the tenth day of life. A similar conclusion was reached by Rodríguez-Blanco et al. [12], who compared the NT-proBNP level at 48–96 h of life and 5–10 days after birth in preterm infants ≤32 weeks of gestation (n = 110). The authors demonstrated that the relationship between NT-proBNP concentration at 48–96 h of life and development of BPD or death is weak and probably confounded by the presence of hsPDA. However, the NT-proBNP level at 5–10 days of life could reliably predict BPD or death independently from the presence of hsPDA (OR = 3.36, 95% CI: 1.52–7.4, p = 0.006). The NT-proBNP concentration > 3348 pg/ml at 5–10 days of life was the best cut-off value for predicting BPD/death with 82% sensitivity and 83% specificity (AUC = 0.87; 95% CI: 0.79–0.95) [12]. According to the results of Harris et al. [28], even lower NT-proBNP concentration > 1598.3 pg/ml on the tenth day of life had 84% sensitivity and 75% specificity in predicting severe BPD (AUC = 0.83) in infants < 30 weeks of gestation (n = 51). At the same time, in a study by Zhou et al. [29] the serum NT-proBNP concentration > 2002.5 pg/ml on the first days of life was predictive for the development of moderate/severe BPD or death with 87.5% sensitivity and 74.7% specificity (AUC = 0.85, 95% CI: 0.79–0.91). Therefore, currently available evidence demonstrates that serum NT-proBNP concentration in very preterm infants in the first 10 days of life could reliably predict the development of BPD or death with high sensitivity and specificity regardless of the presence of PDA.

The advantages of our study are the prospective design and randomized treatment of very preterm infants with PDA diameter > 1.5 mm in the first 3 days of life, that allowed to eliminate the subjective factor in making management decisions, and objectively assess the predictive value of systemic NT-proBNP concentrations for the development of BPD or death in infants with PDA. Our results also indicate that early NT-proBNP screening could reduce the workload and resource utilization associated with PDA assessment in very preterm infants, especially in settings with limited access to around-the-clock echocardiography expertise, and significantly limit the presence of subjectivity component within that assessment. The estimated time required to perform an echocardiographic assessment of the patient significantly exceeds the duration of blood sampling. Additionally, the cost of echocardiographic examination is 10 times higher than the cost of NT-proBNP level determination. Our study did not involve any specific analysis of the economic side of this problem, but obtained data may suggest that NT-proBNP screening could reduce the overall financial burden associated with PDA screening.

Our study also has some limitations, primarily due to its non-blinding design, specific approach to treatment (early administration based only on the PDA size), relatively small number of patients, and proportion of the most immature infants. Prospects for further research are to study the possibility of using serum NT-proBNP concentrations in the first days of life to form a risk group and targeted testing of potential preventive interventions.

In very preterm infants with PDA > 1.5 mm at the age of 24–48 h, serum NT-proBNP concentration could reliably predict the development of BPD or death, regardless of the persistence of PDA, with the highest diagnostic value at 8–9 days. The measurements ≥17,745 pg/ml at the age of 24–48 h had 55% sensitivity and 81% specificity for predicting BPD/death, and concentrations ≥3537 pg/ml on the 8–9 days of life – 60 and 89%, respectively.

Regardless of later death or development of BPD, the serum NT-proBNP values at 2–3 days of life were higher in very preterm infants with hsPDA, and at 8–9 days of life – in babies with persistent PDA.

hsPDA was significantly more often diagnosed in infants who later developed BPD or died, however, pharmacological treatment of hsPDA did not reliably reduce the incidence of BPD or death.