Early Detection of Pulmonary Vascular Dysfunction in Neonatal Bronchopulmonary Dysplasia

Bronchopulmonary dysplasia (BPD), a paramount morbid lung disease typically complicating premature birth, is arguably one of the most vexing clinical problems in the neonatal ICU (NICU). Although the disease is characterized by truncated parenchymal growth, both vascular and alveolar, the pathogenic mechanisms involved remain poorly understood, and the early and late pulmonary complications, particularly development of pulmonary hypertension (PH), are ominous, as they limit survival. Although the definition of BPD is based simply on oxygen dependence at 36 weeks postmenstrual age (PMA), as agreed at a 2001 NIH workshop (1), several pre- and postnatal factors have been linked to its development, including identification of early disrupted pulmonary vascular growth in the form of pulmonary vascular disease (PVD) or PH (2, 3). In fact, these factors may have more relevance to later respiratory complications in early childhood than BPD itself (4) and have led investigators to question the usefulness of the NIH definition of BPD (5). In addition, what has become clearer over the past few years is that early identification of high-risk preterm infants is important for prognostication and possibly early therapy.

In this issue of the Journal, Critser and colleagues (pp. 73–82) tested the hypothesis that neonatal cardiac magnetic resonance imaging (MRI) correlates with BPD severity and can predict short-term clinical outcomes, including the need for PH therapies (6). Building on their previous work showing that an MRI scoring system, based on high- and low-signal intensity lung parenchyma, can detect quantifiable BPD structural abnormalities (7), the investigators retrospectively analyzed a mixed cohort of 52 infants with various degrees of BPD severity who underwent MRI between 39 and 47 weeks’ PMA on a neonatal-sized, NICU-sited 1.5T MR scanner. MR left ventricular eccentricity index (MR-EI), main pulmonary artery to aorta diameter (PA/AO) ratio, and pulmonary arterial blood flow were determined, among several other parameters. Similarly, echocardiograms were obtained, and both MRI and echo indices were compared with BPD severity. Clinical outcomes to be tested against these measurements included NICU and hospital length of stay (LOS), duration of respiratory support, and need for respiratory support at discharge and for PH therapy. The results indicated that, after controlling for confounding factors, MR-EI was associated with LOS and duration of respiratory support, and both increased PA/AO and MR-EI were associated with PH therapy during hospitalization and at discharge. The authors, however, found no correlation between pulmonary arterial blood flow and BPD outcomes. They concluded from their study that specific MRI parameters can provide important clues regarding cardiac morphology that correlate with disease severity and clinical outcomes in neonates with BPD. They specifically highlight that PA/AO ratio and MR-EIs > 1.3 could be used for prognostication for short-term outcomes and potentially for initiation of therapy when the clinical context is appropriate.

This study deals with a challenging neonatal clinical problem (i.e., BPD), which carries a heavy burden in terms of morbidity and mortality. Despite the relatively small size of patients, the authors were able to reach several important conclusions and opened a novel field of imaging with prognostic relevance for a disease that has long been difficult to manage, from both a diagnostic and a therapeutic standpoint. Importantly, the investigators demonstrate that MRI is feasible in a NICU setting and appears to be more reliable than echocardiography in detecting the severity of BPD and its short-term clinical outcomes. The performance of simple imaging biomarkers such as the PA/AO ratio and MRI-EI appears compelling, and their potential clinical use would be quite novel and clinically relevant in this patient population. However, the questions not clearly answered, owing probably to the retrospective nature of this study, are whether MRI is 1) useful for the diagnosis of PH or PVD (a common and serious complication of BPD) or just a predictor of outcomes, and 2) indeed superior to the more cost-effective echocardiography, the preferred modality for PVD diagnosis in high-risk infants. The authors, however, state that comparing the two imaging modalities was not the purpose of their study. In addition, the analysis is limited to relatively short in-hospital outcomes, such as LOS, duration of respiratory support, and initiation of PH therapy. Mourani and colleagues (8) recently demonstrated that early evidence of PVD in preterm infants, based on simple echocardiographic criteria (i.e., one of the following: systolic pulmonary arterial pressure > 40 mm Hg by tricuspid jet velocity, PA/systemic systolic pressure > 0.5, cardiac shunt with bidirectional or right-to-left flow, or any degree of pressure flattening) obtained at 7 days and 36 weeks PMA in a large (i.e., 221 subjects) population of preterm infants, can be detected. They further showed that PVD, in combination with other perinatal factors such as gestational diabetes or mechanical ventilatory support, was a strong predictor for late respiratory disease during childhood (up to 24 mo of age), defined as a diagnosis of asthma or reactive airways disease, a physician diagnosis of bronchiolitis or pneumonia, or hospitalization for a respiratory illness.

Although the study by Crister and colleagues opens a new area of diagnostic investigation in this debilitating disease, it has several notable limitations that were acknowledged by the authors, including the retrospective nature of the analysis, the lack of similar comparator measurements in a group of healthy neonate control subjects (comparisons were made to healthy neonates and neonates with mild BPD), and the relatively small and heterogeneous population of patients from a single center (6). In addition, the diagnosis of PH and, therefore, the decision to start therapy was based on a combination of clinical information and echocardiographic evaluation. This suggests that some patients may have been overdiagnosed, considering the relative lack of accuracy of echocardiography, whereas in others PVD may have been missed if the imaging study was unrevealing (i.e., false negative, although less likely). Right heart catheterization for confirmation of PVD is rarely performed in this vulnerable population, as opposed to adults for whom this test is necessary for the diagnosis. For this study, the authors used evidence of PH based on echocardiographic criteria previously defined by Mourani and colleagues for PVD (8). Other notable limitations include the fact that, although MRI and echocardiographic images were analyzed by readers blinded to the patient identity, there were no attempts at checking accuracy and reproducibility of the imaging parameters (PA/AO and MR-EI) with inter- and intraobserver measurements.

In closing, Crister and colleagues should be commended for pioneering a new avenue of investigation for the early detection of preterm infants at risk for a challenging disease with often somber outcomes (6). However, although this study proposes interesting novel imaging parameters as potential tools for early diagnosis and prognostication, these findings should clearly be validated in prospectively designed, large multicenter studies in which short- and long-term outcomes should both be evaluated.